[00:00:54] <richlowe> on the mailman stuff, I'm getting duplicates of april's mail to osol-arc, (and seemingly only april's)
[00:00:56] <Gman> man, what gives...we announce, um, a black box, and our share price goes down :(
[00:01:01] <gisburn> Gman: and the "future ARC" should work how ? Illuminati-style meetings ? BugZilla bugs ? Snail mail ? Great roundtable ?
[00:01:12] *** axisys has quit IRC
[00:01:35] <Gman> gisburn, there was some internal discussion on how the arc mail filters work, and what mechanism would be best to use
[00:01:48] <sahafeez> !seen benr
[00:01:50] <Drone> benr is currently online in #opensolaris and last spoke on Mon 16 Oct 2006 23:23 GMT, saying 'hey'.
[00:01:50] <sahafeez> hum
[00:01:59] <gisburn> !seen satan
[00:02:00] <Drone> satan (satan!n=chatzill at p54AE763C dot dip.t-dialin.net) was last seen in #opensolaris on Fri 03 Mar 2006 16:21 GMT, saying 'buh!'.
[00:02:16] <benr> bleh.
[00:02:18] <Gman> can we configure drone to msg people instead?
[00:02:19] <benr> hey Gman
[00:02:22] <richlowe> Gman: mail + call is good, it just needs both halfs to actually work.
[00:02:37] <gisburn> Gman: stop bullying Drone.
[00:02:38] <richlowe> Gman: /ignore *!*bevinbot@* works pretty well
[00:02:42] <gisburn> Gman: thanks!
[00:02:47] <Gman> richlowe, nod
[00:02:55] <Gman> richlowe, i'll talk to bevin about changing it :)
[00:03:28] <gisburn> Gman: Drone does a good job of archiving the traffic here and the !seen feature is usefull, too.
[00:03:31] <Gman> richlowe, not getting duplicate mails fwiw
[00:03:45] <Gman> gisburn, yes, it is useful, just doesn't need to print out in the channel
[00:03:47] <richlowe> Gman: I think it's actually a separate mail.
[00:03:57] <richlowe> Gman: probably tried to Roland's problem.
[00:04:05] <Gman> ok
[00:05:11] <gisburn> Gman: I disagree with the Drone thing. Sometimes it is usefull because other people can comment on other people's absence.
[00:05:50] <gisburn> Gman: and the bickering about one line is IMO not worth the issue
[00:06:23] <gisburn> Gman: and complaining to the drone owner will only have one result (as usual): #opensolaris gets removed from drones config
[00:06:33] <gisburn> That happened for other channels, too.
[00:08:31] <gisburn> steleman: ping!
[00:10:36] <gisburn> Anyone else wants to bully Drone ?
[00:10:44] * gisburn waves with an axe...
[00:12:08] *** nwf has quit IRC
[00:14:12] <gisburn> good.
[00:17:23] <gisburn> stevel: do you speak chinese ?
[00:17:45] <gisburn> stevel: any idea what "dhlang" means ?
[00:20:31] * gisburn can't find any code for the ldom support...
[00:21:41] <gisburn> Is the ldom support already in the OpenSolaris tree ?
[00:26:12] <richlowe> yes.
[00:27:07] <Doc> it is, but last i looked (a few weeks ago) the BIOS needed to support it wasnt out yet
[00:27:22] <Doc> well, technically the hypervisor version needed
[00:29:32] <mdj> Any significant advantage to running DS 5.2 vs. OpenLDAP for centralized authentication? (Assuming you are not trying to integrate the JES Identity stack?)
[00:32:26] * gisburn reads http://en.wikipedia.org/wiki/Discworld_gods to learn more about the "dhlang" ...
[00:39:12] * gisburn growls...
[00:39:22] <gisburn> alanc_work: ping!
[00:39:43] <gisburn> alanc_work: is m64's 8+24bit mode broken in B48 ?
[00:40:27] *** eboutilier has quit IRC
[00:42:40] *** mdj has quit IRC
[00:43:38] *** Tpenta has joined #opensolaris
[00:44:40] *** ChanServ sets mode: +o Tpenta
[00:46:53] *** xzilla is now known as xzi11a
[00:50:45] *** stevel changes topic to "Latest releases: SXCR: 50, ON build: 50, ON nightly: 20061016"
[00:51:07] <stevel> cd: http://javashoplm.sun.com/ECom/docs/Welcome.jsp?StoreId=7&PartDetailId=Sol-Express_b50-x86-SP-G-B&TransactionId=try
[00:51:12] <stevel> dvd: http://javashoplm.sun.com/ECom/docs/Welcome.jsp?StoreId=7&PartDetailId=Sol-Express_b50-DVD-x86-SP-G-B&TransactionId=try
[00:51:31] * gisburn wants B51
[00:51:33] <richlowe> and this has mustang as the default java, right?
[00:51:34] <gisburn> NOW!
[00:51:51] *** jcea has quit IRC
[00:51:56] <stevel> i thought that was 51
[00:52:45] <richlowe> stevel: jek's heads-up said it was a respin of 50.
[00:52:46] <richlowe> that's why I'm asking. :)
[00:52:50] <stevel> ah
[00:52:53] <stevel> then 50, yes :)
[00:54:05] <richlowe> ... now I'm more puzzled than when I started.
[00:55:07] <stevel> oh, btw, minimum build required to build has now been bumped up to snv_46
[00:55:12] <stevel> starting with today's drop
[00:55:38] <Tpenta> whew. it's a good thing that I have 46 on the machine I just kicked of fthe encumbered bins build on :)
[00:57:02] <Kush-> how would i go about getting the contents of a <%cleanup> component and displaying it else where in the template/compoentn
[00:57:13] <richlowe> stevel: yeah, I was going to say it could be worth noting that in the announcement.
[00:57:47] <Kush-> re that profiler, i am trying to do a c.start=time.clock() in base.py and then in the autohandler do a nother  end=time.clock() inside <%cleanup>
[00:57:56] *** movement has quit IRC
[00:58:09] <richlowe> but then my day turned into a joyous progression of repeated X crashes.
[00:58:27] *** ShadowHntr has quit IRC
[00:58:32] <alanc_work> gisburn: no clue about 8+24
[00:59:48] *** solarisjon has quit IRC
[01:00:10] <Kush-> damn, wrong channel. :|
[01:02:19] *** apa has joined #opensolaris
[01:03:54] *** lloy0076 has joined #opensolaris
[01:04:09] *** movement has joined #opensolaris
[01:04:21] *** lloy0076 has left #opensolaris
[01:08:50] <Error_404> does SNMP allow you to on/offline machines?
[01:08:55] <Error_404> or anything for that matter?
[01:09:25] *** koolniczka has left #opensolaris
[01:09:27] <jamesd> it shouldn't
[01:10:12] <Error_404> oh, n/m... wake on lan should work
[01:10:41] <Error_404> for the record, i have no intention of actually doing anything with this knowledge, it's just a thought experiment
[01:10:52] <Error_404> as in, how would a load balancer bring machines online as it needs them
[01:12:37] <Error_404> so instead of wasting power on 6 machines to service 6 requests/s, have one machine service enough to get 90% in use before onlining another box
[01:13:48] <sommerfeld> well, there are remote power switching boxes which can be controlled via SNMP
[01:14:01] <apa> so if the load balancer goes down , the entire system is unavailable ?
[01:14:06] <sommerfeld> (apc masterswitch is one which I know you can use snmpset to toggle a power switch)
[01:15:18] <apa> how do you in general, solve the one point failure problem ?  or is the only true way to have the client have a list of servers it can switch to as needed ?.
[01:15:45] <twincest> apa: vrrp/hsrp and various other ip failover systems are good for this particular problem
[01:15:49] <sommerfeld> apa: throwing money at the problem.
[01:16:08] <sommerfeld> two load-balancers in a active/standby configuration
[01:16:47] <apa> ok , do how does the client know which load balancer to use ?  ie what IP ?.
[01:17:05] <Error_404> load balancer heartbeat, i'd assume
[01:17:16] <twincest> apa: vrrp and hsrp share a single IP between the devices
[01:17:17] <Error_404> as load balancer A dies, balancer B assumes the IP & continues
[01:17:21] <twincest> apa: when one fails, the other takes the IP over
[01:17:22] <apa> or it all boils down to that there is still  a one point failure anyhow ?
[01:17:29] <apa> ah oki.
[01:17:42] <apa> nice
[01:17:58] <apa> thanks alot for explaining
[01:19:02] *** lloy0076 has joined #opensolaris
[01:19:45] <Error_404> are there already facilities to do what i was talking about, btw?
[01:19:48] <twincest> there are other ways to do it.  for udp services (dns mainly) you can use anycast, for example
[01:20:40] <Error_404> for turning on machines as the already in service ones become overloaded?
[01:21:40] <lloy0076> Wish me luck.
[01:21:52] <lloy0076> Hopefully I come back intact after that BFU.
[01:21:53] <twincest> error: i would have thought constantly powering machines off and on was worse than just leaving them on
[01:22:00] <twincest> error: it's not good for the hardware
[01:22:19] *** lloy0076 has left #opensolaris
[01:22:54] <Error_404> well, it wouldn't fare well for the lifetime of the hardware, but i'm thinking pure electricity & cooling cost
[01:24:35] <apa> its bad for all electronics, its current surges during start etc.  and esp the HDs take a beating, ex: the fat in the bearings is cold and causes extra wear at start.
[01:25:21] <Error_404> okay okay, so it was a bad idea
[01:25:25] <apa> the hardware cant go to low power mode etc ?.
[01:25:41] <apa> well its doable, just not good from the hardware point of view =)
[01:26:04] <Error_404> depending on what hardware, one may not really care
[01:26:31] <Error_404> a pile of beige boxes are disposable, more or less
[01:26:42] <apa> oki. then problem is solved.
[01:27:01] <Auralis> i would config it so that the system are in lowpower/power saving mode, and have scripts running on that the load of the active systems, if a certain threashold is crossed upwards new systems get brought up from powersave, if crossed down, systems get taken back to powersave
[01:27:56] <Auralis> and possible a shared storage as well
[01:28:40] <apa> well most modern cpus do the power saving by them self if configed in bios and or driver
[01:28:41] <Error_404> Auralis: that's what i was thinking
[01:29:32] *** Haggi_8-] has joined #opensolaris
[01:29:40] *** lloy0076 has joined #opensolaris
[01:30:15] *** slowhog has joined #opensolaris
[01:30:46] <lloy0076> Amazing; it worked.
[01:31:00] <Error_404> damn, and here i thought i stumbled on a brilliant idea to save money
[01:34:47] *** jcea has joined #opensolaris
[01:35:00] *** Kush- has quit IRC
[01:35:58] <lloy0076> Error_404: Unfortunately, most ideas to make lots of money and/or save lots of money are already known.
[01:37:32] *** lloy0076 has left #opensolaris
[01:43:06] *** deather_ has joined #opensolaris
[01:51:40] *** deather has quit IRC
[02:04:25] *** bengtf has quit IRC
[02:17:47] *** steleman has quit IRC
[02:19:26] *** steleman has joined #opensolaris
[02:20:34] *** McBofh has joined #opensolaris
[02:22:50] *** sahafeez has quit IRC
[02:22:59] *** john5 has joined #opensolaris
[02:24:32] <gisburn> alanc_work: ping!
[02:24:50] <gisburn> !summon kupfer
[02:24:53] <gisburn> !summon kupfer
[02:25:14] * gisburn wishes Drone would send people a pager notification or SMS for !summon
[02:25:34] <richlowe> me too, I think what they did to you for disturbing them that way would be *hilarious*. :)
[02:25:45] * gisburn could kick people out of their beds at 3:04am
[02:25:55] * stevel could kick gisburn anytime of the day
[02:26:00] <stevel> ;-)
[02:26:08] <stevel> it's 5:30pm. past normal business hours. he's probably gone home
[02:26:38] *** sommerfeld has quit IRC
[02:33:44] <alanc_work> gisburn: pong
[02:33:55] <alanc_work> was down the hall trying to debug xorg bugs
[02:36:19] *** Haggi_8-] has quit IRC
[02:37:49] <gisburn> alanc_work: see invitations
[02:38:07] <alanc_work> oh, I hadn't seen those
[02:39:04] <slowhog> alanc_work: the X src on opensolaris is for which NV build?
[02:39:38] <slowhog> fontconfig works for me. :-)
[02:39:41] <alanc_work> I don't rememeber which was last released - the changelogs on opensolaris.org show the mappings to nevada builds
[02:39:59] <alanc_work> have had to cut back on the opensolaris X releases due to lack of resources
[02:41:28] <slowhog> it has a date 0914, but don't see the mapping
[02:42:00] <slowhog> when will you guys move to svn or mercurial?
[02:42:07] <slowhog> ;-)
[02:43:08] <slowhog> 0914 maybe 43?
[02:43:12] *** steleman has quit IRC
[02:43:24] *** steleman has joined #opensolaris
[02:43:55] <slowhog> oh, I see the link, it's 49
[02:53:07] <alanc_work> slowhog: never
[02:53:36] <alanc_work> at least, that's the current plan - plans are always subject to change
[02:54:36] <slowhog> not even the open part?
[02:57:55] *** piwi has joined #opensolaris
[02:58:14] <alanc_work> nope
[02:58:27] <alanc_work> any sort of migration requires work
[02:58:58] <alanc_work> we're overbooked working on the code as it is - no time for distractions like changing code management tools
[03:09:22] *** Kernel86|Laptop has joined #OpenSolaris
[03:17:33] <gisburn> What does " Kudos, Roland" mean ?
[03:17:46] <alanc_work> "Good Job Roland"
[03:17:50] * gisburn prepares a shotgun
[03:17:53] <gisburn> oh
[03:18:00] <alanc_work> "Yay Roland!"
[03:18:01] * gisburn throws rhe shotgun away
[03:18:09] <Error_404> Roland as in Edirol?
[03:18:18] <gisburn> uhm
[03:18:40] <Error_404> as in, roland 303
[03:18:42] <alanc_work> http://en.wikipedia.org/wiki/Kudos
[03:18:44] * gisburn prepares a double-barrel 40mm Bossler frontloader shotgun and aims it at Error_404
[03:18:46] <Error_404> the musical insturment company
[03:18:49] <gisburn> Error_404: how did you call me ?
[03:19:31] <Error_404> what?
[03:19:50] <alanc_work> if you're asking about Dan's mail, he is agreeing that a default of gmacs is much better than a default of none
[03:20:17] <alanc_work> don't worry, be happy!
[03:21:33] <Gman> ooh, gnome system tools has been approved by lsarc
[03:21:57] * Gman remembers the time a few years ago we were too scared to even suggest integrating it ;)
[03:24:48] <Error_404> i've nothing but respect for roland... they make some of the best midrange music production equipment on the market
[03:25:05] <richlowe> Hey Gman.
[03:25:09] *** eugene has left #opensolaris
[03:25:16] *** Xh4 has joined #opensolaris
[03:25:47] <Gman> hey hey richlowe
[03:25:47] <alanc_work> why does libtool have to suck so badly?
[03:26:00] <Gman> heh
[03:26:05] * Gman feels alanc's pain
[03:26:41] <Gman> gisburn, nitpick 'usability'
[03:26:57] <gisburn> Gman: ?!
[03:27:14] <Gman> that's the proper spelling
[03:27:17] <dvorak> alanc_work: unfortunately most other things suck more
[03:27:35] *** benr has quit IRC
[03:27:57] <dvorak> I actually wouldn't be entirely unhappy with libtool if it had a halfway working concept of multiarch hosts
[03:28:31] <alanc_work> it's sad that rm *.la has to be done so often to make it work
[03:28:46] <Gman> heh
[03:28:52] <Gman> we do that in our build env too
[03:29:01] <Gman> and i think pretty much every linux distro does that too
[03:29:24] <alanc_work> when I started working on the Xorg modular builds last year I hacked up a version of install.sh that just skipped the installation of any *.la files
[03:29:38] <dvorak> Gman: debian and suse definitely don't
[03:30:01] <Gman> hrm, i think debian gnome does
[03:30:06] <Gman> maybe i'm wrong
[03:30:17] <dvorak> I find the .la files usually work, if the contents are actually correct
[03:30:42] <dvorak> getting the contents correct when you're using a separate build root and install root can be a big pain in the ass though
[03:32:20] <Tpenta> THAT explains why my sparc encumbered builds were running so slowly. I only had one board DR'd into the 4800, .... DR's in the other two and hope things improve
[03:32:24] <alanc_work> bah - the only hit google finds for the wierd linker error I'm getting from libtool is for the l10n messages catalog for Sun ld
[03:32:56] <gisburn> heh
[03:33:27] <dvorak> alanc: what's the error?
[03:33:55] <alanc_work> offset: 0x242b: relocation requires reference symbol
[03:34:16] <dvorak> that is odd
[03:34:20] <dvorak> is it c++ code?
[03:34:27] <alanc_work> nope, pure C
[03:34:37] <alanc_work> (and some asm)
[03:35:48] <alanc_work> works if I remove that one .o file from the link, but it's a rather important .o
[03:40:54] <Gman> holy shit, the jds svn migration is well underway
[03:40:57] <Gman> wooohooo!
[03:41:39] <stevel> indeed it is
[03:41:58] <gisburn> stevel: ping!
[03:42:14] <gisburn> stevel: can I include svn.genunx.org to the "SCM page" ?
[03:42:31] <stevel> which SCM page?
[03:42:53] <gisburn> stevel: http://www.opensolaris.org/repository/Project.Display.mtw?project_name=ksh93-integration&id=1133
[03:43:22] <stevel> nope
[03:43:31] <gisburn> stevel: why ?
[03:43:35] <stevel> that's a dynamic page, generated so you can maintain opensolaris.org SCM repositories
[03:43:48] <stevel> there's no way we can run commands like svnlook or svnadmin on a SVN repository that isn't hosted locally
[03:43:49] <Gman> you could move the repo across though
[03:44:30] * stevel nods
[03:47:26] *** slowhog has left #opensolaris
[03:53:50] <Tpenta> stevel: should have the nd stuff for 20061016 shortly (now that I've added the otehr two boards onto the 4800)
[03:54:17] <stevel> great
[03:55:56] <alanc_work> looks like it was the inline asm in that file that confused ld - fortunately it was easily replacable
[03:56:07] <stevel> okay i'm outta here. time to head home
[03:56:08] <stevel> 'night folks
[03:56:22] <Gman> night stevel
[03:56:30] *** stevel has quit IRC
[03:56:35] <Tpenta> g'nite steve
[03:56:59] <Tpenta> damn too slow
[03:57:48] <gisburn> Uhm... seems the /etc/ksh.kshrc ARC case will be another horrible thing.
[03:58:27] <Tpenta> it looked pretty straightforward to me
[03:58:32] <Tpenta> which is why I didn't comment
[03:58:54] *** piwi has quit IRC
[03:59:29] <LeftWing> As long as it contains set -o vi =)
[03:59:29] <gisburn> Tpenta: Originally we thought we can get it approved this wednesday - but the commentary will likely prevent that and require another phone conference.
[03:59:58] <gisburn> LeftWing: sure. If you continue to teach the stutents here how vi works.
[04:00:55] <LeftWing> Throw an "echo For more information about vi, go here: http://www.opensolaris.org/some/sort/of/vi/tutorial.html" into /etc/ksh.kshrc ;P
[04:02:06] <Gman> gisburn, i don't necessarily think that arguments holds for emacs either fwiw
[04:02:08] <gisburn> umpf
[04:03:10] <richlowe> sodding mit.co.id
[04:05:20] <alanc_work> I would have thought it would be easy, but then richlowe had to start bringing facts to the table
[04:06:19] <Gman> heh
[04:06:38] <Gman> facts, sanity, ...
[04:07:35] <alanc_work> though in truth, asking any group of Unix users, whether ARC or not, to choose between vi & emacs is just begging for a flamewar
[04:08:42] <Error_404> not really anymore
[04:08:43] <Tpenta> It looks to me like the only non-converged issue is the "style" issue, I still have hope that it will converge
[04:08:54] <Error_404> nobody cares, now i'ts gnome vs. kde
[04:09:12] <Tpenta> by the way, by email count, it's still an order of magnitude less of a monster than the initial case ;)
[04:09:26] <icon> not much of a contest there considering gtk's wild popularity
[04:09:39] <Tpenta> and I am pleased to see civility reigning
[04:09:45] <richlowe> there's been no argument at all regarding the presence of the file, beyond the somewhat messy spray of shell configuration in /etc
[04:09:48] <LeftWing> GNOME is certainly improving.
[04:10:03] <alanc_work> GNOME vs. KDE is easy - Sun picked GNOME, and it's too painful to change Solaris now
[04:10:17] <icon> gnome has suprised me greatly, instead of bloating to large degrees like most other largish env's, its stayed pretty tight
[04:10:34] *** slowhog has joined #opensolaris
[04:10:35] <LeftWing> Heh, it's pretty bloated.  But it feels snappier at 2.14 than it did at 2.4
[04:10:50] <icon> yes and no :)
[04:10:51] <Gman> gnome is rad!
[04:11:02] <icon> gnome has stayed very focused on what they do
[04:11:06] * Gman wearing his big gnome foot tshirt :)
[04:11:09] <Error_404> my experience is just that if you connect more users to it, it starts to chew through memory in ways that kde doesn't
[04:11:12] <icon> kde on the otherhand is well... kde :)
[04:11:21] <LeftWing> Error_404: You mean CDE? ;P
[04:11:22] *** slowhog has quit IRC
[04:12:04] <icon> imho, kde spends more time trying to win the windows folks over instead of focusing on a unix desktop like gnome has
[04:12:31] <icon> im a developer, i could care less if i have a 'control panel' ;)
[04:12:34] <LeftWing> I do wish GNOME had more configurables.
[04:12:46] <icon> well, the linux hal stuff is quite impressive
[04:13:02] <icon> as much as i dig on linux, thats one thing that was done very well
[04:13:14] *** slowhog has joined #opensolaris
[04:13:17] <LeftWing> And we'll have it soon. ;P
[04:13:20] <icon> here here :)
[04:13:35] <icon> id also like to see the security type support that linux has too
[04:13:53] <alanc_work> why should we weaken Solaris security down to the Linux level? 8-)
[04:13:56] <icon> ie: if i run dhcpadm etc. i get prompted for a super user login
[04:14:08] <icon> bah its just convenience
[04:14:15] <icon> i get annoyed with having to setenv DISPLAY all day long :D
[04:14:21] <alanc_work> oh that - that's part of the gnome system tools work, isn't it Gman?
[04:14:32] <LeftWing> gsudo? ;P
[04:14:35] *** yongsun has joined #opensolaris
[04:14:40] <alanc_work> gtksu?
[04:14:53] <Gman> gksu
[04:14:54] <Gman> yep
[04:15:00] <icon> ahh cool
[04:15:06] <icon> that would be really nice to see :)
[04:15:18] <icon> oh speaking of which, the new theme knocks me dead
[04:15:19] <Gman> it'll be in b53
[04:15:32] <icon> now *that* is a solaris desktop l&f :D
[04:15:33] <LeftWing> What's the new theme?
[04:15:34] <slowhog> alanc_work: ln -s /etc /usr/etc fix my fontconfig problem with b50
[04:15:40] <icon> i cant remember the name offhand
[04:15:42] <Gman> nimbus
[04:15:45] <icon> thats it
[04:15:48] <richlowe> it'd be nice if there were nimbus icons too.
[04:15:52] <Gman> there are
[04:15:54] <richlowe> or if it defaulted to tango :)
[04:15:56] <alanc_work> slowhog: /usr/etc ?
[04:15:57] <Gman> the latest drop has them
[04:15:58] <icon> Gman: which release were those added?
[04:16:00] <icon> ahh cool
[04:16:07] <Gman> they're pretty fab too
[04:16:08] <richlowe> Gman: nice.
[04:16:08] <slowhog> yes
[04:16:25] <alanc_work> weird
[04:16:28] <Gman> please file bugs on missing icons though
[04:16:37] <alanc_work> I'd tell the fontconfig guy, but he doesn't work here anymore 8-(
[04:16:42] <Gman> [sad thing is that the designer who did them has moved to apple]
[04:16:51] <icon> Gman: which icon did you guys go with for the menu button?
[04:17:11] <Gman> it's still the orange S shaped curve
[04:17:15] <Gman> with the globe
[04:17:24] <Gman> [which will be a coffee cup in solaris]
[04:17:25] <icon> ahh damn, i was hoping the java logo would stick :)
[04:17:29] <icon> cool
[04:17:45] <icon> so i guess you guys are still sticking with the debranding in opensol then?
[04:17:47] <slowhog> when I build  fontconfig yesterday, I build with default confdir without specifying --with-confdir=/etc/fonts, it complains and ask to have /usr/etc/fonts
[04:17:53] <slowhog> so I make a symbolic link
[04:17:54] <Gman> icon, yeah, we kind have to
[04:17:56] <richlowe> Gman: do they happen to be in osol/jds/extras/?
[04:18:04] <Gman> no reason we can't change the world icon to something else though
[04:18:05] <icon> Gman: oh well, nothing a quick mv wont fix :)
[04:18:14] <Gman> richlowe, should be part of SUNWgnome-themes
[04:18:35] <icon> honestly, i really like the simplicity the interface has been heading the last year or so
[04:18:37] <Gman> richlowe, i can upload a package for you if you like
[04:18:47] <icon> its much more pleasant than jds 1 and 2 were
[04:19:03] <Gman> yeah, much nicer
[04:19:08] <Gman> hal will make a big difference too
[04:19:10] <Triskelios> soooo much nicer
[04:19:15] <icon> when will hal make it in?
[04:19:20] * slowhog likes my clearlooks2-squared theme. :-)
[04:19:20] <Gman> it's already in
[04:19:21] <richlowe> it did already.
[04:19:23] <Gman> so, b51
[04:19:25] <richlowe> b51
[04:19:26] <richlowe> damnit.
[04:19:27] <icon> oh wow
[04:19:33] <icon> time to update ;)
[04:19:43] <icon> when is the next sxcr due out then? (im lazy)
[04:20:05] <Gman> 2 weeks time i guess
[04:20:08] <Tpenta> given 50 came out in teh last 24 hours, about two weeks
[04:20:11] <Gman> i think derek just announced b50 today
[04:20:15] <icon> gotcha
[04:21:27] <icon> Gman: have you been reading discuss much the last week or so?
[04:21:46] <Gman> desktop-discuss? or opensolaris-discuss?
[04:21:51] <icon> opensol
[04:21:56] <LeftWing> Does gnome-pilot sit on top of pilot-link?
[04:22:01] <slowhog> minutes ago, I removed the /usr/etc/ link and rebuilt fontconfig with --with-confdir=/etc/fonts, and that will cause segment fault
[04:22:02] <Gman> i summarize it each monday :)
[04:22:05] <Gman> LeftWing, yes
[04:22:13] <icon> Gman: im the (un)fortunate ports guy :)
[04:22:29] <alanc_work> desktop-discuss has been more fun with the sparc-graphics-group-bashing 8-)
[04:22:33] <Gman> icon, yep, did a whois a couple minutes ago
[04:22:38] <Gman> good luck with that
[04:22:40] <icon> ahh groovy
[04:22:43] <slowhog> so I switch back the b50 libfontconfig, and ln -s /etc /usr/etc, and that, fix fontconfig segment fault for me. :-)
[04:22:58] <icon> well, ive been talking with laca alot, and were thinking of using pkgbuild for the build system
[04:23:10] <icon> i was curious how the jds guys would feel about making jds ports based
[04:23:25] <icon> would at least make updates way easier for you all
[04:23:29] <alanc_work> slowhog: I never would have thought to try that, and I can't begin to explain it
[04:23:42] <alanc_work> but it's time for me to go home now, so I won't even try
[04:23:45] *** alanc_work is now known as alanc-away
[04:23:54] <Gman> icon, would be useful, though i'd obviously hesitate until it actually got into mainstream solaris
[04:24:05] <icon> *nod*
[04:24:22] <icon> still trying to figure out a bloody language to do this thing in
[04:24:27] <icon> not much to pick from in order to keep it pure
[04:24:32] <Gman> C of course ;)
[04:24:41] <icon> well that was my first thought
[04:25:03] <Gman> sounds like you can't leverage a lot of code?
[04:25:08] <icon> but well... sol 8 is still kicking, so that means sun4m, sun4u, x86
[04:25:11] <icon> well yes and no
[04:25:16] <icon> more like leveraging ideas :D
[04:25:23] <Gman> you could target nevada+
[04:25:31] <icon> id like that
[04:25:33] <Gman> and avoid older platforms
[04:25:47] <icon> most ports systems are based off of -CURRENT with hacks for backwards compatiblity
[04:25:52] <icon> it works pretty well for the most part
[04:25:56] *** slowhog has quit IRC
[04:26:44] <icon> i still wonder a bit if scripting would be better
[04:26:53] <icon> which leaves just tcl and python (perl is dead to me)
[04:26:56] <richlowe> beyond spec files, pkgbuild and the wrapper I forget the name of would seem to do nearly everything in that regard.
[04:27:08] <icon> it gets about 80% of the way there
[04:27:15] <icon> theres a lot thats lacking there
[04:27:23] <icon> no real repository support
[04:27:41] <icon> things like smf scripting, organization, indexing etc.
[04:27:58] <richlowe> indexing and such would possibly need adding, yeah.
[04:28:07] <icon> it needs to be a cohesive system
[04:28:17] <icon> we have the chance to do it right the first time, so thats what the plan is :)
[04:28:26] <richlowe> the thing is, there's two views (at least) of a ports system.  There's "a convenient way to build packages, which people then install", and "Have people build everything themselves, but maybe provide packages too"
[04:28:49] <icon> well this will bridge that gap a bit
[04:29:04] <icon> dports takes the stance that when a package is built, an actual package is generated
[04:29:15] *** SYS64738 has joined #opensolaris
[04:29:18] <icon> the bsd's portage, etc. actually build straight to the target in the fs, so no packages are created
[04:29:31] <icon> sysv4 would be used to manage all packages, dependencies etc.
[04:29:41] <richlowe> The thing with Solaris is, you can't necessarily assume the system has a compiler. :)
[04:29:45] <Gman> i don't think i want to see solaris becoming another gentoo though :)
[04:29:54] <icon> god no
[04:30:04] <richlowe> so as a mechanism to package software, I like it, as something intended for users to directly use in the common case, I'd have concerns.
[04:30:07] <icon> i detest what gentoo has done for ports systems
[04:30:48] <icon> the ports system is also a way to actually create packages for common distribution
[04:31:22] <icon> it just depends on your target audience. sun proper obviously has a more vested interest in packages they control
[04:31:35] <icon> however, thats incredibly inconvenient for power users
[04:31:56] <icon> the bsd's have used the mechanism for many years with much success
[04:32:07] <icon> usage is braindead simple
[04:34:44] *** slowhog_home has joined #opensolaris
[04:45:39] *** slowhog has joined #opensolaris
[04:45:42] <Gman> http://www.gnome.org/~gman/SUNWgnome-themes.tar.gz
[04:45:46] <Gman> http://www.gnome.org/~gman/SUNWgnome-themes-devel.tar.gz
[04:45:53] <Gman> for anyone who wants newer packages for nimbus
[04:46:07] <Gman> [built against gnome 2.16, but should probably be compatible]
[04:46:18] <icon> nice!
[04:46:21] <icon> thanks gman
[04:46:42] <Gman> it's what is up on dlc.sun.com fwiw - just extracted the relevant packages
[04:47:10] <icon> nah its easy for me, i tend to be a bit lazy on keeping jds up to date
[04:47:25] <icon> i think im still on 46
[04:48:22] <SYS64738> Is it dual p3 SMP supported on nevada ?
[04:48:41] <SYS64738> I am trying to install but it reboots
[04:51:55] *** delewis has quit IRC
[04:55:30] *** haltonAFK is now known as halton
[04:57:15] <icon> wow
[04:57:19] <icon> archive.org can be pretty fun
[04:57:59] <icon> found an old screenshot of my sparc in college ;)
[05:02:25] *** yongsun has quit IRC
[05:04:15] *** jwtodd has joined #opensolaris
[05:09:04] *** Xh4 has quit IRC
[05:12:12] *** Gman is now known as GmanAKF
[05:12:20] * icon &
[05:17:06] *** jsubl2 has joined #opensolaris
[05:18:22] *** yongsun has joined #opensolaris
[05:20:02] *** yongsun has quit IRC
[05:20:09] *** yongsun has joined #opensolaris
[05:22:08] *** karrotx has joined #opensolaris
[05:27:58] *** SYS64738 has quit IRC
[05:30:15] *** ShadowHntr has joined #opensolaris
[05:31:02] *** qbit has joined #opensolaris
[05:31:18] *** lisppaste3 has quit IRC
[05:33:50] *** lisppaste3 has joined #opensolaris
[05:37:11] *** dclarke has joined #opensolaris
[05:37:27] <dclarke> hello
[05:38:01] <dclarke> testing gaim 2.0.0 here
[05:40:13] <dclarke> hello ?
[05:40:18] <dclarke> this think working ?
[05:40:32] * dclarke wanders to another channel
[05:46:30] <dclarke> :-/
[05:47:14] *** dclarke has left #opensolaris
[05:49:39] *** jsubl2 has quit IRC
[05:54:45] *** Symmetria has joined #opensolaris
[05:55:05] <Symmetria> lo all
[05:55:27] <Symmetria> Gawd I need sleep, 5:55am is *NOT* a time to be awake
[05:55:47] <ShadowHntr> lol
[05:55:50] <ShadowHntr> true.
[05:56:45] <Symmetria> heh GEANT network doing maintenance which is gonna require work on my routers
[05:57:21] <Symmetria> and heh, while normally I'd let some other appie do this config, cant afford anyone screwing up that particular connection
[05:57:27] *** dclarke has joined #opensolaris
[05:57:46] <Symmetria> 2 things which are the life blood of our organization, our peer to Internet2 and our peer to geant
[05:57:46] <dclarke> looking for signs of life here ...
[05:57:50] *** logic has joined #opensolaris
[05:58:00] <noyb> hi
[05:58:03] <dclarke> hello
[05:58:07] <noyb> happy now?
[05:58:08] <noyb> :)
[05:58:09] <ShadowHntr> hehehehe
[05:59:59] <noyb> dclarke: I'm interested to see if you can reproduce a bug in gaim.  Gman tells me it's fixed at the trunk, and will be out in snv_53, but I'm curious if you could try something for me.
[06:00:11] <Symmetria> heh man I cant wait to move my mirror server onto solaris but ugh I cant move it right now cause of having to travel this weekend
[06:00:18] <dclarke> sure ..
[06:00:27] <Symmetria> heh today I start the copying of data though from the old server to the new server
[06:00:29] <noyb> dclarke: offline
[06:00:35] <dclarke> offline ?
[06:00:40] <noyb>  /msg me
[06:00:44] <Symmetria> anyone here got any idea how long its gonna take to copy 3.8 TB over gigE :p
[06:02:56] <Auralis> about 380 seconds if your boxes can keep up
[06:03:28] <Symmetria> errr Auralis no
[06:03:32] <Symmetria> gigE = gigabit
[06:03:39] <Symmetria> TB = terabyte
[06:04:07] <Symmetria> its actually 380*8 IF it ran at exactly gigE (but point is that will never happen)
[06:04:24] *** lloy0076 has joined #opensolaris
[06:04:37] <Symmetria> heh, drives, memory, everything cannot keep up with that :)
[06:04:41] <Symmetria> hence the question
[06:05:36] *** slowhog_home has quit IRC
[06:06:28] <Symmetria> oh and knock off 56 bytes per frame so if you're running 9000 MTU jumbo frames ......... :p
[06:09:00] <rydis> What sort of disks are there? For commodity hardware, that'd probably be the limiting factor.
[06:11:03] <Symmetria> rydis heh large raid 5 iSCSI arrays
[06:11:22] <Symmetria> those are SATA 500gig 16meg cache western digitals
[06:12:18] <Symmetria> heh working on an average of 34meg/second Im looking at 32 hours
[06:12:25] <Symmetria> *shrug* not *THAT* bad
[06:12:30] *** logic_ has quit IRC
[06:14:04] <McBofh> Symmetria: assuming you are comparing it to using carrier pigeons :)
[06:15:24] <Symmetria> heh McBofh *shrug* copying that kinda data without incredibly high speed disks and huge amounts of ram caching and either infiniband or 10gE nics
[06:15:25] <Symmetria> sux :p
[06:15:31] <Symmetria> cause its always gonna take an age
[06:15:45] <dclarke> I have a question about ZFS compared to SVM
[06:15:51] <dclarke> and its an easy one
[06:16:24] <dclarke> essentially .. is there a way to provide RAID 0+1 and RAID 1+0 in both ?
[06:16:51] <dclarke> so .. given a stack of disks in two arrays on two controllers
[06:17:16] <dclarke> create a mirror of one disk from one array with a disk from the other array
[06:17:21] <Symmetria> dclarke cant answer that unfortunatly, all my raids are done in hardware
[06:17:24] <dclarke> repeat until all disks are mirrored
[06:17:31] <dclarke> then stripe
[06:17:44] <dclarke> hardware raid .. fine
[06:18:10] <Symmetria> dclarke heh that should be possible, the closest I've ever com to doing that is striping 2 raid-6 arrays
[06:18:24] <dclarke> wtf is RAID 6 ?
[06:18:25] <Symmetria> but again, the raid-6 arrays were iSCSI SAN's in hardware
[06:18:26] *** eokyere has left #opensolaris
[06:18:36] <Symmetria> dclarke basically raid 5 with 2 redundant disks
[06:18:50] <dclarke> oh .. so RAID 5 plus hot spares
[06:18:52] <Symmetria> its raid 5 + hot spare
[06:18:53] <Symmetria> yeah
[06:19:00] <dclarke> to me .. thats just RAID 5
[06:19:27] <dclarke> trust me .. software RAID 5 is a _big_ mistake
[06:19:46] <Symmetria> yeah I dont run software raid 5, as I said, those were hardware san's
[06:26:08] <McBofh> dclarke: for read/write loads, yes. for guaranteed read-only loads, not so bad
[06:26:18] <McBofh> hw raid5 still kicks sw raid5's arse though
[06:26:55] <dclarke> every time
[06:26:58] <dclarke> yep .. I agree
[06:27:01] <Symmetria> heh I run all my raids on LSI MegaRaid's
[06:27:33] <Symmetria> internal raid runs LSI MegaRaid and the iSCSI SAN's run LSI MegaRaid as well, then Qlogic iSCSI HBA Offload to reach the SAN as well
[06:27:45] <icon> *yawn*
[06:28:48] <dclarke> I have to deal with some A5200's here
[06:28:55] <dclarke> and a deck of disks in each
[06:29:30] <Symmetria> man Im getting pissed at geant
[06:29:38] <dclarke> a pile of SEAGATE-ST3146807LC-0007
[06:30:08] *** lloy0076 has left #opensolaris
[06:31:00] <gdamore> hi *
[06:31:07] <icon> hrmm
[06:31:10] <icon> how is dri support going so far?
[06:35:19] *** halton is now known as halton_meeting
[06:41:16] <gisburn> Does anyone know a way to turn the buffering for a tty channel off ?
[06:41:25] <gisburn> (from the shell level)
[06:41:35] <icon> hrmm
[06:41:50] <gdamore> use the stty command.
[06:42:07] <icon> cant find the flat he needs though...
[06:42:09] <gdamore> i have a hack somewhere that used stty and dd to get single character immediate prompting in a ksh based menu.
[06:42:21] <icon> s/flat/flag/
[06:42:30] <icon> bah i really need to get vm's setup on this machine
[06:46:01] *** Method has joined #opensolaris
[06:46:01] *** jcea has quit IRC
[06:46:09] <richlowe> stty raw/stty -raw, I think.
[06:46:42] <icon> hrmm
[06:46:51] <icon> that doesnt affect buffering does it?
[06:47:08] <icon> just how it processes input/output
[06:47:09] *** laca has quit IRC
[06:53:35] *** Symmetria has quit IRC
[06:54:31] *** MadMethod has quit IRC
[06:57:26] *** jerome__ has joined #opensolaris
[06:57:34] <jerome__> Hi there
[06:57:43] <jerome__> bash-3.00# showmount -e localhost
[06:57:43] <jerome__> showmount: localhost: RPC: Program not registered
[06:57:45] <jerome__> any clues
[06:57:48] <gdamore> icon: buffering _is_ how it processes input.
[06:57:57] <jerome__> bash-3.00# ps -ef | grep rpcbi | grep -v grep
[06:57:57] <jerome__>   daemon   243     1   0 14:45:20 ?           0:00 /usr/sbin/rpcbind
[06:58:06] <gdamore> make sure you have filesystems exported in /etc/dfstab
[06:58:14] <gdamore> and then make sure nfsd is running.
[06:58:36] <gdamore> on solaris 9 and earlier it was /etc/init.d/nfs.server start.  I don't knwo the exactl SMF incantation off hand.
[06:58:48] <jerome__> still not werkin
[06:59:03] <jerome__> bash-3.00# ./nfs.server stop
[06:59:04] <jerome__> bash-3.00# ./nfs.server start
[06:59:04] <jerome__> bash-3.00# showmount -e localhost
[06:59:04] <jerome__> showmount: localhost: RPC: Program not registered
[06:59:04] <gdamore> ps -ef | grep nfsd
[06:59:23] <gdamore> it won't do anything if you don't have exports listed in /etc/dfs/dfstab
[06:59:24] <jerome__> it seems that nfsd is not willing to start...
[06:59:40] <jerome__> ok
[06:59:41] <jerome__> let me see
[06:59:47] <McBofh> jerome__: running s10 or solaris express?
[07:00:07] <McBofh> either way, you need to ensure that nfs/server and nfs/client services are running
[07:00:13] <gdamore> in this case it shouldn't matter, although the nfs.server start is certainly  pre-s10 thing.
[07:00:15] <jerome__> running build 49
[07:00:39] <McBofh> gdamore: true, true
[07:00:42] <gdamore> i'm not entirely sure that you need the client running to see the server.
[07:00:50] <jerome__>   daemon  2985     1   0 22:00:38 ?           0:00 /usr/lib/nfs/nfsd
[07:00:54] <jerome__> bash-3.00# showmount -e localhost
[07:00:54] <jerome__> showmount: localhost: RPC: Program not registered
[07:00:56] <jerome__> hmmmm
[07:00:59] <Error_404> i'm making quiche
[07:01:00] <Error_404> yum. quiche
[07:01:28] <gdamore> what's in dfstab?
[07:01:38] <jerome__> bash-3.00# cat /etc/dfstab
[07:01:42] <jerome__> /zfs
[07:01:58] <gdamore> that isn't good syntax.
[07:02:09] <jerome__> ah okay
[07:02:11] <jerome__> hold on
[07:02:23] <gdamore> it should contain share -F nfs commands. :-)
[07:02:26] <jerome__> okay
[07:02:28] <jerome__> let me fix
[07:02:47] <jerome__> share -F nfs -o ro -d "zfs" /zfs
[07:02:48] <jerome__> ~
[07:03:02] <gdamore> after fixing, do /etc/init.d/nfs.server stop; /etc/init.d/nfs.server start.  apparently it is still there (the script) on s10
[07:03:06] <jerome__> ok
[07:03:28] <jerome__> works thanks man
[07:03:29] <McBofh> jerome__: if that's a zfs rather than ufs filesystem, set the sharenfs property
[07:03:37] <jerome__> sharenfs?
[07:03:50] *** Jiko has quit IRC
[07:03:51] <McBofh> jerome__: man zfs
[07:03:52] <richlowe> zfs set sharenfs=<options> /that/file/system
[07:04:12] <jerome__> richlowe: where is this going to?
[07:04:44] <McBofh> jerome__: richlowe, gdamore and I are trying to get you to do the correct thing for zfs rather "what we used to do with ufs"
[07:04:45] <jerome__> ok
[07:05:01] <jerome__> what should go in options?
[07:06:17] <jerome__> got it
[07:06:20] <jerome__> let me try
[07:11:12] *** lloy0076 has joined #opensolaris
[07:11:48] <lloy0076> Every time I "su -" my PATH gets set to /usr/sbin:/usr/bin...
[07:12:36] <lloy0076> ...so I have a 'fix_paths.sh' which I eval to get my path back to what I want it to.
[07:15:14] *** McBofh has quit IRC
[07:20:40] <richlowe> su - will cause it to take the path root has set, yes.
[07:21:56] <lloy0076> Indeed, but I have a different SUPATH set in /etc/default/login
[07:22:07] <lloy0076> (which has about 3 more searches than just those two)
[07:24:12] *** delewis has joined #opensolaris
[07:25:32] *** Jiko_ has joined #opensolaris
[07:38:11] <Tpenta> non-debug encumbered-binaries available for 20061016
[07:48:00] *** gisburn has quit IRC
[07:53:10] *** yongsun has quit IRC
[07:53:19] *** yongsun_ has joined #opensolaris
[07:59:18] <asyd> \_o<
[08:07:02] *** lloy0076 has left #opensolaris
[08:08:26] <richlowe> Tpenta: nice :)
[08:10:31] <richlowe> Hm, java 1.5 appears to still be the default in snv_50.
[08:10:44] <richlowe> I guess the the b50 that's SXCR currently isn't the respin referenced in the headsup
[08:13:38] <delewis> SXCR isn't using b50, I think.
[08:13:45] <richlowe> yes, it is.
[08:13:55] <delewis> the gate close was 10/02
[08:14:27] <delewis> b50 is 20061009, isn't it?
[08:14:51] <richlowe> the gate closed 10/02, it was delivered to the WOS 10/09, today is 10/18, and SXCR is available.
[08:15:04] <richlowe> well, it's only just 10/18. :)
[08:16:02] <richlowe> 20061009 would be a snapshot midway through b51.  b50 is b50, 20061002 is a snapshot as of the gate closing.
[08:16:30] <delewis> oh, so anything say on genunix.org that uses a timestamp such as yyyymmdd is a nightly build?
[08:16:48] <delewis> and actual bxx releases don't correspond to nightlies?
[08:16:49] <richlowe> it's the source as it was at about 11pm PST on that date.
[08:16:54] <richlowe> well 'pacific'.
[08:16:55] <delewis> right
[08:17:00] <delewis> alright, that makes sense.
[08:18:24] <richlowe> gah.  When liveupgrade gets done, I'm going to sync SS11 to every BE on this machine, and not blow them away with an incautious liveupgrade for the 3rd time this week.
[08:19:01] <delewis> hehe, Live Upgrade is great for ON builds.
[08:19:48] <delewis> I've got an SXCR BE that is continously upgraded to SXCR releases and a BE that contains a BFU'd SXCR.
[08:20:25] <richlowe> yeah, then you drop the right compilers onto the bfu'd BE.
[08:20:39] <richlowe> and upgrade the SXCR, but you can't sync from the BFu'd BE that you're running, so you boot to the SXCR be, sync.
[08:20:42] <richlowe> and blow away the compilers.
[08:20:52] <richlowe> in my case, you do that twice in succesion.
[08:20:54] <richlowe> :)
[08:21:29] <delewis> are you still transitioning between SS10 and SS11 or something?
[08:21:31] *** Netwolf has joined #opensolaris
[08:21:42] <richlowe> nope, I just don't upgrade SXCR that often.
[08:21:47] <delewis> ah
[08:21:58] <delewis> me either -- if I didn't I wouldn't have much time to do anything else.
[08:22:08] <delewis> I'm still running snv48 :-)
[08:30:05] *** bengtf has joined #opensolaris
[08:43:32] <clee> does osol have ata-over-ethernet support yet?
[08:46:32] <clee> I know that the coraid guys released a driver for it
[08:46:50] * clee doesn't know exactly how to find out if it's in the community releases, either
[08:51:40] *** Sieghard has joined #opensolaris
[08:52:26] <Sieghard> i have discovered the strangest solaris problem in my entire life
[08:55:12] *** bengtf has quit IRC
[09:01:02] *** Cyrille has joined #opensolaris
[09:03:25] <jerome__> how do you nfs export with noatime?
[09:03:40] <jerome__> share -F nfs -o ro,noatime -d "zfs" /zfs
[09:03:43] <jerome__> right?
[09:05:44] <jerome__> any clues?
[09:11:49] <Sieghard> maybe it's an import option?
[09:12:25] *** mazon is now known as Mazon
[09:12:36] <jerome__> sorry you're right
[09:12:40] <jerome__> I'm just tired
[09:18:26] *** slowhog has left #opensolaris
[09:19:02] *** bengtf has joined #opensolaris
[09:23:01] <richlowe> great, more spam via Jive.
[09:23:49] <Sieghard> if i copy these files over the network to the solaris server, the machine will freeze for a few seconds and the ethernet links will go down/up
[09:24:05] <Sieghard> why does it only happen to these files?  what's so special about them
[09:24:23] <Sieghard> and why is the problem so intermittent?
[09:26:52] *** benr has joined #opensolaris
[09:26:52] *** ChanServ sets mode: +o benr
[09:32:35] <Tpenta> evening mr rockwood
[09:32:44] *** deather_ is now known as deather
[09:44:00] *** dunc has quit IRC
[09:45:51] *** Dar_HOME is now known as Dar
[09:49:35] <SymmWork> hrm
[10:01:34] *** lloy0076 has joined #opensolaris
[10:02:44] *** ShadowHntr has quit IRC
[10:10:04] *** GmanAKF has quit IRC
[10:10:06] *** alanc has quit IRC
[10:10:46] *** lloy0076 has quit IRC
[10:10:50] *** alanc has joined #opensolaris
[10:14:29] *** lloy0076 has joined #opensolaris
[10:16:02] *** damienc has joined #opensolaris
[10:17:25] *** Tpenta has quit IRC
[10:19:36] *** dunc has joined #opensolaris
[10:30:11] *** damienc has quit IRC
[10:31:10] <timeless> what's the zonecfg syntax for setting a inherited-pkg-dir ?
[10:31:13] *** damienc has joined #opensolaris
[10:31:42] *** Gman has joined #opensolaris
[10:34:05] <trygvis> I would guess: add inherited-pkg-dir
[10:34:43] <timeless> that gave me "usage: ..."
[10:34:51] <timeless> which i think basically means "invalid syntax"
[10:35:39] <trygvis> strange, works for interrffaces/
[10:35:44] <trygvis> interfaces
[10:37:48] <timeless> it works for net too
[10:37:51] <timeless> but that doesn't help me :)
[10:38:06] <timeless> anyway, nexenta in zone has dpkg
[10:38:09] <ofu> hmmm, i have an x4200, 2 CPUs and 8 memory modules. Why do these modules have to be put into the white sockets?
[10:38:15] <timeless> but it doesn't seem to have apt or ldd
[10:38:37] <ofu> the paintings on the chassis say, i can use white or black ones
[10:38:46] <lloy0076> Heh - Sun recommends a 120Hz processor for x86 platform. I would really hate to see it run on such a system even with a stack of RAM.
[10:39:15] <timeless> i think i remember people talking about laptops w/ that kind of cpu
[10:39:26] <timeless> i used a p133laptop for a while, but it wasn't running solaris
[10:39:45] <timeless> it's not bad if you're running a text editor, ssh, and something like vnc
[10:39:50] <lloy0076> I'm just trying to imagine using Gnome2 on such a beast; it's slow enough as it is.
[10:39:52] <timeless> rdpclient/whichever
[10:40:03] <timeless> "don't  go there"
[10:40:06] *** deather has quit IRC
[10:40:07] *** peteh has joined #opensolaris
[10:40:12] *** deather has joined #opensolaris
[10:40:19] <timeless> 120mhz w/ a good pipe is fine as a thin client
[10:40:35] <lloy0076> Fair enough.
[10:42:51] <lloy0076> That said, I take it that the performance characteristics of the UltraSparc IIIi processors at 1.34Gz compete with AMD Opteron Model 1218 dual core chips...
[10:43:18] <lloy0076> ...at least the pricing of tope end Sun 20 M2s vs Sun 25 low ends seems to indicate that.
[10:43:31] <lloy0076> <-- isn't familiar with UltraSparc performance characteristics
[10:43:55] * timeless isn't a hardware person
[10:44:08] <lloy0076> heh
[10:44:46] <timeless> i just use low and high end hardware and know limitations :)
[10:45:43] <lloy0076> I'm running it on an AMD XP 2400+ with 512Mb of RAM; Solaris runs all right on that.
[10:46:09] <lloy0076> It would be nice if Java Swing Apps (or SWT Apps) ran faster but alas!
[10:51:50] <timeless> trygvis: ok, after getting rid of inherit, it turns out i do have apt-get
[10:52:04] <timeless> so all i need to do is figure out how to plumb my network / zone
[10:52:52] <trygvis> _now_ you can use add net :P
[10:55:51] <timeless> http://www.opensolaris.org/os/project/crossbow/CrossbowRelease08-2006/GettingStarted/;jsessionid=5AE831FC6B0CA37D17690F6BAE839D38
[10:56:02] <timeless> is that page a bad starting point?
[10:56:07] <timeless> # dladm create-vnic -d e1000g0 -m 0:1:2:3:4:5 -b 10000 2
[10:56:08] <timeless> dladm: unknown subcommand 'create-vnic'
[10:56:37] <trygvis> hm, why would you need to do that? isn't that only if you want trunking and stuff?
[10:56:47] <trygvis> I always just add net and get a new virtual interface
[10:56:50] <timeless> because i really don't know what i'm doing :)
[10:56:55] <trygvis> you can also do: ifconfig bge0:1 plumb
[10:57:00] <trygvis> and then configure bge0:1
[10:57:29] <timeless> for the time being i want the zone to share the ip of the global zone
[10:57:43] <timeless> and just have client access (i.e. no local services)
[10:58:04] <trygvis> you can't do that, it requires it's own interface and ip
[10:58:23] <trygvis> we just give it local (10.0.0.x) addresses and access it through the global zone
[10:59:31] <timeless> i presume bge0 is a specific vendor?
[11:00:03] <trygvis> it's broadcoms gigabit driver
[11:00:18] <timeless> # ifconfig e1000g0:1 plumb
[11:00:22] <trygvis> you can add virtual interfaces like that for all nics
[11:00:23] <timeless> do i really want to do this?
[11:00:25] <trygvis> yep
[11:00:43] <trygvis> but it should be e1000:1
[11:00:47] <timeless> if the nexenta zone just wants to talk to the global zone
[11:00:56] <trygvis> the driver is called bge, thus bge0
[11:01:02] <timeless> does it actually send things over the wire?
[11:01:19] <timeless> swift% /sbin/ifconfig -a4|grep e100
[11:01:19] <timeless> e1000g0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
[11:01:19] <timeless> e1000g0:1: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
[11:01:23] <trygvis> that depends on your routing
[11:02:15] <timeless> can i depend on solaris to do my routing for me (correctly)? :)
[11:03:02] <lloy0076> lol
[11:03:11] <lloy0076> Are you asking whether Solaris will root your routes up?
[11:03:29] <timeless> root in that context would mean fsck
[11:03:34] <timeless> no, i don't want it to do that :)
[11:04:07] <lloy0076> You could always just try it. If you can keep talking here -- assuming you're on the same connection -- it's fine, if you disappear, something went wrong!
[11:04:20] * timeless chuckles
[11:04:47] <timeless> zonecfg:nexenta:net> set physical=e1000g0:1
[11:04:48] <timeless> e1000g0:1: physical interface name required; logical interface name not allowed
[11:05:01] * timeless grumbles
[11:06:46] *** MattMan has joined #opensolaris
[11:06:55] * lloy0076 sigh
[11:07:03] <timeless> ok, so, do i basically create :1 and :2
[11:07:17] <lloy0076> I am trying to work out whether I should release a bunch of PHP code under the CDDL, LGPL or a Revised BSD license.
[11:07:20] <lloy0076> It's a nightmare.
[11:07:23] <timeless> have nexenta use :1 w/ say 192.168.20.101
[11:07:32] <timeless> and have global use :2 w/ say 192.168.20.1 ?
[11:07:51] <timeless> release it under a severable any of these licenses thing
[11:08:15] <timeless> but seriously, unless you have a reason, i'd go w/ the bsd license
[11:08:37] <lloy0076> We want to make sure that any changes to the code actually go back to the community.
[11:08:55] <timeless> ah
[11:08:56] <lloy0076> We don't care if it's used proprietary or if it links with Bill Gate's Wonderful Monopoly.
[11:09:05] *** bengtf has quit IRC
[11:09:28] <lloy0076> I'm not interested in making BG's monopoly mine (a la GPL), just in making sure that BG doesn't enhance our code and not give the enhancements back.
[11:10:04] <lloy0076> [Ideally, I want a license where only enhancements and improvements are permitted to be given back! LOL]
[11:10:15] <timeless> you don't want people to add bugs?
[11:10:35] <timeless> that's so politically incorrect!
[11:10:53] <timeless> debian reserves the right of its users to add bugs to your code
[11:10:54] <timeless> :)
[11:10:54] <lloy0076> So, Revised BSD -- whilst it makes the codebase go open source -- doesn't really site well with me for what I want to do with this code.
[11:11:03] * timeless nods
[11:11:12] <timeless> for that i personally like mpl
[11:11:19] <lloy0076> Hence, I'm kind of stuck with the LGPL or the CDDL or may MPL.
[11:11:33] <timeless> if you care about people linking w/ gpl code you multilicense w/ lgpl/gpl
[11:11:48] <timeless> i read cddl a while ago, but i can't remember it
[11:12:02] <andersmo> mmmm... the smell of legal spaghetti!
[11:12:05] <timeless> i think it's probably somewhere along the lines of mpl, but please please don't quote me
[11:12:14] <lloy0076> timeless: It's based on MPL.
[11:12:30] <lloy0076> The glitch is that we don't like the GPL at all.
[11:12:31] <timeless> oh good :)
[11:12:36] <timeless> heh
[11:12:42] <timeless> that's ok, GPL doesn't like MPL
[11:12:49] <Cyrille> or any other license.
[11:12:58] <lloy0076> We suspect that users of the code will want to link it against proprietary stuff.
[11:13:07] <lloy0076> Which kind of gives the GPL the boot.
[11:13:21] <kimc> give it the boot :)
[11:13:24] <timeless> MPL was of course designed to allow linking to priorietary
[11:13:31] * timeless is all in favor of giving gpl the boot
[11:13:45] <lloy0076> As far as I'm concerned, the GPL has a purpose but it reeks of: "We want everything for nothing.".
[11:13:59] <lloy0076> I fail to see how FREE software as in beer feeds anyone.
[11:14:00] * lloy0076 sigh
[11:14:11] <timeless> anyway, i've made my views known, but right now i'd rather figure out this routing mess
[11:14:12] <lloy0076> Sorry, I got on my soapbox.
[11:14:13] <soultan> anybody good with hardware?
[11:14:16] <Cyrille> isn't there something like LGPL which allows linking with proprietary bits but not changing the code itself without releasing changes?
[11:14:17] <soultan> i'm trying to install an amd opteron 64 heatsink on an elitegroup motherboard. however there's some huge ati metal object standing in the way, making it seem impossible to hook the heat sink down to its cpu mount. anybody know why?
[11:14:18] <timeless> am i right about needing 3 ifs?
[11:14:21] <soultan> i don't want to jack that metal object off incase it's not meant for that
[11:14:42] <timeless> cyrille: if the proprietary bits are part of the "os"
[11:14:45] <timeless> whatever the fsck that means
[11:14:59] <soultan> timeless: it's ok to say fuck
[11:15:09] * timeless should have said 'heck'
[11:15:12] <kimc> it also smells of "..We're so righteous we almost can't stand ourselves..'
[11:15:13] <soultan> well that's what my teacher tells me
[11:15:17] <lloy0076> soultan: No idea, sorry. Removing bits off mother boards sounds dodgy though.
[11:15:18] <andersmo> soultan: timeless may not think so? =)
[11:15:28] <timeless> i'm kinda old school
[11:15:39] <timeless> for instance, i think the idea of a girl telling a guy 'f-ck you''
[11:15:40] <lloy0076> timeless is a timeless was.
[11:15:41] <timeless> is stupid
[11:15:51] <lloy0076> s/a/as/
[11:15:53] <Cyrille> "heck"? what is that? Is that the place you're darned to if you don't believe in gosh?
[11:15:55] <timeless> it's an open invitation ...
[11:16:03] <soultan> true logic
[11:16:05] <soultan> err
[11:16:07] <soultan> true lloy0076
[11:16:10] <andersmo> And after a fixing a few broken file systems, fsck is as good a swear word as any.
[11:16:36] <timeless> answers.com claims heck is in the dictionary
[11:16:39] <lloy0076> Anyway, I shall sojourn to my local eating place for dinner!
[11:17:07] <kimc> do the useage rights to fsck carry the gpl ?
[11:17:12] <timeless> adjourn, no?
[11:17:48] <lloy0076> Sojourn and adjourn.
[11:18:01] <lloy0076> It involves both a journey and an adjournment...
[11:18:10] <lloy0076> Anyway, until later!
[11:18:11] *** lloy0076 has quit IRC
[11:18:20] <timeless> so, back to networking :)
[11:23:24] <timeless> stupid question
[11:23:52] <timeless> what would cause my keyboard to not result in repeated keys when i hold down a key?
[11:25:16] <timeless> ah, gnome prefs were silly :)
[11:26:13] *** Abyte-wiser has joined #opensolaris
[11:30:45] <timeless> e1000g0:1: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
[11:30:46] <timeless>         inet 192.168.20.1 netmask ffffff00 broadcast 192.168.20.255
[11:30:51] <timeless> swift% /usr/sbin/ping 192.168.20.1
[11:30:54] <timeless> ^C
[11:31:03] <timeless> did i do something wrong? :(
[11:32:42] *** sgorilla78 has joined #opensolaris
[11:35:28] *** bengtf has joined #opensolaris
[11:37:39] <mustang> <timeless> e1000g0:1: flags=1000842<BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
[11:37:46] <mustang> you're missing the magic word
[11:37:51] <mustang> "UP"
[11:37:52] <mustang> :)
[11:38:22] <timeless> thanks
[11:38:35] *** pjd_ is now known as pjd-
[11:38:40] <timeless> ok, my zones can see eachother :)
[11:39:29] *** apa has quit IRC
[11:40:38] <timeless> add net default: gateway 192.168.20.1: insufficient privileges
[11:41:19] <asyd> the routing table can be edited only from the global zone
[11:41:20] <asyd> s10
[11:41:21] <asyd> oups
[11:41:57] *** bengtf has quit IRC
[11:43:35] *** kimc has left #opensolaris
[11:44:25] *** coffman has joined #opensolaris
[11:48:32] *** bengtf has joined #opensolaris
[11:51:22] *** Gman is now known as GmanZZZ
[11:52:22] <timeless> ok
[11:52:25] <timeless> that didn't work correctly
[11:52:28] *** phalenor1 has joined #opensolaris
[11:52:29] <timeless> my box just rebooted
[11:53:18] <timeless> does snv_49 normally give panic messages when it dies?
[11:53:56] <sgorilla78> hello
[11:54:07] <sgorilla78> is opensolaris production ready?
[11:54:13] <sgorilla78> what dist would you recommend
[11:54:24] <sgorilla78> someone sent me some links to zfs, and it seems really cool
[11:54:42] <sgorilla78> i am going to be setting up a backup system tommorow, want to see what options i have
[11:55:05] <coffman> sgorilla78: u could use solaris 10 or express
[11:55:48] <sgorilla78> solaris 10 with xfs by default?
[11:56:00] <timeless> xfs?
[11:57:01] <PerterB> whoa, flashback to IRIX
[11:57:17] *** phalenor- has quit IRC
[11:57:54] <PerterB> sgorilla78: Solaris 10 6/06 has zfs, yes... but not as the default filesystem type if that's what you mean
[11:58:04] <sgorilla78> PerterB: oh
[11:58:27] <sgorilla78> is zfs being used in production now?
[11:58:33] <coffman> sure
[11:59:08] *** trs81 has quit IRC
[11:59:21] <coffman> solaris 10u3 should be come any time now
[12:00:18] <sgorilla78> cool
[12:00:24] <sgorilla78> how much is it for commericial use?
[12:00:44] <coffman> 100%?
[12:01:11] <trygvis> solaris for production usage is free
[12:02:00] <sgorilla78> ok im download it now
[12:03:05] * timeless frowns
[12:03:15] <coffman> sgorilla78: if u need patches, u can buy a service plan for like 120$ a year - security patches are for free
[12:03:25] <sgorilla78> oh cool, that is reasonable
[12:03:36] <coffman> its pretty cheap
[12:03:46] <sgorilla78> i just need a decent backup server
[12:03:53] * timeless frowns
[12:03:57] <sgorilla78> going to be putting rsync and openvpn
[12:04:14] <sgorilla78> i have a bunch of disks that are different size
[12:04:23] <timeless> i don't understand why my system forgot its ip address and similar settings
[12:04:26] <sgorilla78> seems like zfs is a better solution than RAID
[12:04:31] * timeless needs to make sure it doesn't forget this time
[12:04:39] <hali> timeless: what do you mean with "forget"?
[12:04:50] <timeless> hali: it booted but didn't know its static ip
[12:04:57] <timeless> which i had configured when i installed the system
[12:05:03] <hali> timeless: how did you set it? the correct way or just with ifconfig?
[12:05:03] <timeless> or at least, i think i did
[12:05:15] <timeless> i really don't remember, my memory is bad :)
[12:05:22] <hali> so it seems
[12:06:28] <hali> timeless: try setting the hostname in /etc/hostname.INTERFACENAME, the default route in /etc/defaultrouter etc
[12:08:11] <timeless> is there a simple way (not rebooting) to test that a given set of those are correctly written to disk?
[12:08:30] <timeless> can i ifconfig INTERFACE down/up or similar to have it pull the settings from the right files?
[12:09:06] <PerterB> yup
[12:09:47] <ofu> svcadm network/physical restart?
[12:09:58] <ofu> never tried it, but it might work
[12:10:11] *** calumb has joined #opensolaris
[12:11:04] <coffman> sgorilla78: atm there is only tun for openvpn on solarsi
[12:11:24] <PerterB> hmm, that's a point... if you just unplumb your primary interface in sol10 (eg to test this) does SMF get upset and re-plumb it?
[12:13:27] *** TBCOOL has quit IRC
[12:18:45] <timeless> ok, i have /etc/hostname.e1000g0 and it contains 'swift'
[12:19:01] <timeless> /etc/hosts has a line 10.30.31.32    swift
[12:19:17] <timeless> but when i do svcadm restart svc:/network/physical:default
[12:19:21] *** sgorilla78 has quit IRC
[12:19:40] <timeless> e1000g0: becomes inet 0.0.0.0 --> 172.21.50.1
[12:20:15] <timeless> hrm, ok, maybe the hosts line is more like 172.21.50.something :)
[12:25:07] <timeless> hrm, is http://blogs.sun.com/jwadams/entry/debugging_smf_5_managed_processes full of oforeign blog spam? :(
[12:26:24] <PerterB> check you don't have a different entry in /etc/inet/ipnodes
[12:27:33] *** benr has quit IRC
[12:28:59] <timeless> /etc/hosts -> ./inet/hosts
[12:29:02] <timeless> erm
[12:29:14] <PerterB> yeah, that's expected
[12:29:36] <timeless> no differences between inet/hosts and inet/ipnodes
[12:32:56] <timeless> ok, one of the logs says
[12:33:09] <timeless> configuring IPv6 interfaces:ifconfig: swift: bad address
[12:33:13] <timeless>  e1000g0.
[12:33:49] <timeless> and then it tried dhcp
[12:35:43] *** yongsun_ has quit IRC
[12:40:27] *** Abyte-wiser has quit IRC
[12:41:02] *** dunc has quit IRC
[12:42:56] *** coffman_ has joined #opensolaris
[12:43:24] *** medar has joined #opensolaris
[12:46:27] <timeless> ok, killing the dhcp file for my interface seems to have fixed it :)
[12:46:50] <timeless> but pinging 'swift' from the nexenta zone still causes my system to reboot
[12:46:55] <timeless> not quite ideal
[12:49:28] <PerterB> definitely not optimal
[12:49:43] <timeless> where do i look? :)\
[12:50:01] <PerterB> is there a panic message and crash dump?
[12:50:10] <timeless> no, it just reboots
[12:50:20] <timeless> i hear one tone and see messages from bios
[12:50:39] <timeless> i'm used to panic messages and crash dumps and whatever
[12:51:08] <PerterB> and it only happens from the nexenta zone?
[12:51:25] <timeless> i don't have many zones :)
[12:51:33] <timeless> and i certainly haven't tried playing w/ networking before
[12:52:12] <PerterB> just wondering whether it's a hardware problem is all
[12:52:17] <timeless> if i were to place a bet, it'd be related to the fact that i'm asking it to use 192.168.20.101 to ping the 172. address
[12:52:36] <timeless> and with all 3 of those addresses (192.168.20.1) being on the same nic
[12:52:56] <PerterB> that should work ok
[12:54:52] <timeless> well, i can't ping the nameserver from the zone
[12:55:25] <timeless> and i can ping the zone's name, the globalzone's name, and localhost
[12:55:54] <timeless> and i can't ping the global zone's router
[12:56:03] *** coffman_ has quit IRC
[12:59:15] *** coffman has quit IRC
[12:59:17] <trygvis> timeless: paste your setup on rafb.net/paste
[12:59:27] <trygvis> I think you might be where I am now with my zone setup
[13:00:44] <timeless> ok, what do you need?
[13:01:12] <timeless> i think i'm going to stick up all of my files onto a web server instead of a pastebin
[13:01:16] <trygvis> the interfaces, their op and netstat -r -n
[13:01:29] <trygvis> s,op,ip,
[13:04:21] <timeless> http://www.webwizardry.net/~timeless/osol/
[13:04:23] <timeless> gimme a bit
[13:06:26] <trygvis> right, you seem to have a similar setup to me
[13:06:29] *** peteh has quit IRC
[13:07:27] <timeless> ok, i've put up zones, the get_nexenta script, and my settings
[13:09:45] <timeless> oh
[13:09:56] <timeless> i bet i want to make nexenta share /var/crash/cores :)
[13:09:57] <trygvis> what you need is a nat setup
[13:09:58] <trygvis> or routing
[13:10:03] <timeless> yeah
[13:10:10] <trygvis> let me know when you figure it out :)
[13:10:14] <timeless> heh
[13:10:51] <timeless> hrm
[13:11:01] <timeless> coreadm won't tell me what signal triggered a core?
[13:11:31] *** lloy0076 has joined #opensolaris
[13:14:53] *** dunc has joined #opensolaris
[13:55:57] <timeless> trygvis: http://opensolaris.org/os/community/documentation/opensolaris_guide/;jsessionid=76192D0C0FC1DE7377D209293F979AD6 ?
[13:57:18] <timeless> (ipf /usr/share/ipfilter/examples/nat-setup)
[13:58:50] *** peteh has joined #opensolaris
[13:59:10] <trygvis> hey, cool
[14:02:59] *** Tpenta has joined #opensolaris
[14:03:22] <ofu> where can i download the companion cd for sol10u2?
[14:04:12] <trygvis> the link is next to the download links for the "main" cds
[14:04:31] *** TBCOOL has joined #opensolaris
[14:06:35] * lloy0076 thwaps self
[14:06:45] <lloy0076> Why am I downloading a Linux binary when I'm using Solaris...
[14:07:04] <ofu> aah, now i found it, thx
[14:08:03] *** lloy0076 has quit IRC
[14:08:43] * timeless is confused
[14:10:52] *** halton_meeting has left #opensolaris
[14:11:30] <ofu> so there is only one companion dvd, where i can find x86 and sparc pkgs?
[14:14:15] <lasseoe> yes
[14:14:24] *** timeless has quit IRC
[14:15:50] *** timeless has joined #opensolaris
[14:15:55] <timeless> oops
[14:16:24] <timeless> are man files supposed to reference themselves?
[14:16:37] <timeless> i.e. should man 9 foopy says see foopy(9) ?
[14:18:29] <lasseoe> if it's a manpage about iterations, then I don't see why not ;-P
[14:18:36] <richlowe> I don't believe so, it'd be a fairly useless thing for them to do.
[14:18:43] <jteo> Easter Egg!
[14:18:55] <richlowe> though "If you want to know more, read this over again" *is* generally true.
[14:18:56] <timeless> man 4 ipf
[14:19:08] <timeless> search for "See ipf"
[14:20:07] <richlowe> ipf(4) only references ipf(1M) and ipfilter(5)
[14:20:13] <richlowe> ipf(1M) only references ipf(4) and ipfilter(5)
[14:20:24] <timeless> hrm
[14:20:33] <richlowe> I suspect you're falling over the man syntax.
[14:20:40] <timeless> ok, so my problem is i typed man 4 ipf and got ipf(1M)
[14:20:40] <richlowe> it needs either -s <section> or <page>.<section>
[14:20:51] <Tpenta> richlowe: I am hoping that the comments I just added to the fast track for /etc/ksh.kshrc might get us to convergance
[14:21:02] <delewis> <section> <page> wouldn't be bad, either :-)
[14:21:06] <timeless> richlowe: sorry, linux lets me do man section page
[14:21:15] <Tpenta> timeless: try man -s4 ipf
[14:21:20] <richlowe> timeless: so do many other things, I do that stuff constantly before I remember.
[14:21:23] <timeless> no excuse, sure, but clearly i haven't done it often / recently on unix
[14:21:26] <delewis> most UNIXes I've worked with support <section> <page>
[14:21:27] <jteo> i would say something mean and how everything isn't Linux, but i try to be polite.
[14:21:31] <delewis> AIX certainly does
[14:21:41] <Tpenta> v7 did
[14:21:51] <delewis> as did BSD4.3, IIRC.
[14:22:00] <richlowe> Tpenta: yeah, I've been wanting to explain what is at least my position on that.
[14:22:10] <richlowe> but I haven't found a way that reads right the 2nd time.
[14:22:19] <timeless> ?
[14:22:28] <Tpenta> my position is that I'm not sure it's actually an issue for the ARC, being an implementation detail
[14:22:31] * timeless tries to read what richlowe wrote right the first time and fails
[14:22:46] <richlowe> Tpenta: I talked this through with sommerfeld (or you, or alanc, or someone) when I first was talking to Roland about it.
[14:22:57] <richlowe> whether the actual interaction with it counts as 'interface'
[14:22:57] <Tpenta> wasnt me
[14:23:14] <richlowe> I somewhat think it does, given how annoyed I would be should that interface change drastically with no warning.
[14:23:27] <richlowe> which is also why I'm not happy about 'gmacs' being the effective default being Uncommited.
[14:23:34] <richlowe> surely if it's *that* desirable, it could be Commited?
[14:23:52] <Tpenta> I would argue the pathname to the file is the interface, the contents thereof are not
[14:24:09] <timeless> tpenta: but the parsing for the file are part of the interface, right?
[14:24:10] <richlowe> the contents thereof affect what is effectively an interface of the utility processing them.
[14:24:13] <timeless> s/are/is/
[14:24:15] <Tpenta> sorry, the pathname to the file and what it is used for
[14:24:19] <richlowe> while I would normally agree, it does make a large change to the behaviour of ksh.
[14:24:27] <richlowe> which I would call interface, I think.
[14:24:38] <Tpenta> parsing the file, ... hmmmm I would argue not part of this case, part of the ksh93 case
[14:24:40] <richlowe> (the *interactive* behaviour, rather, it makes no change to script processing)
[14:25:53] <Tpenta> besides if I can make the "style issue" out of scope for PSARC, the case has converged
[14:26:02] <Tpenta> and will go straight through tomorrow
[14:26:06] <delewis> indeed, and the argument that gmacs would migration for Linux users is moot, considering bash already exists if they want gmacs (they're unlikely to to use ksh93, anyway, as unfortunate as that is)
[14:26:07] *** yydzero has joined #opensolaris
[14:26:11] <delewis> ease*
[14:26:42] <Tpenta> the whoel argument is really moot anyway as any sysadmin who really cares is going to change the default
[14:29:41] <timeless> ok, what magic causes the ifconfig -a4 inet x --> y arrow bit to happen during boot?
[14:29:45] <richlowe> I think there is no solution I would find more acceptable than the current one.
[14:30:04] <timeless> e.g. ifconfig dev 172.16.17.10/24 172.16.17.1
[14:30:09] <richlowe> (the proposed one, rather)
[14:30:09] <timeless> the .1 part
[14:30:18] *** calumb is now known as calLNCH
[14:30:34] <Tpenta> To be honest, I don't think it really matters what the default is. you are going to offend someone. deal with it.
[14:30:47] <Tpenta> implementation detail
[14:30:50] <richlowe> and that brings me to the "explaining my position" part.
[14:30:50] <twincest> how do i remove an fs from a zone? 'remove fs dir=/home' claims 'remove fs: No such resource with that id'
[14:30:55] <richlowe> I don't care about *what* the default is.
[14:30:57] <richlowe> my concern is there being one.
[14:31:25] <richlowe> but I run two ways on that, I don't believe defaulting to 'none' is a nice thing to do.
[14:31:35] <richlowe> but I also don't think adjusting the way things work *now* is nice either.
[14:31:53] <richlowe> since there's no way to satisfy both of those, I think the proposed way is the best way, barring silly arguments about *what* the default is.
[14:32:09] <Tpenta> Throw a d6, halve the number, move on
[14:32:30] * timeless has /etc/defaultrouter but it didn't seem to be enough
[14:32:32] <jteo> i prefer a d24.
[14:32:49] <Tpenta> I *have* seen a d30 ;)
[14:33:08] <richlowe> Tpenta: anyway, I want to see it run through to a call again, 2006/550 was enlightening :)
[14:33:36] * timeless likes d8 personally
[14:33:46] <timeless> d100s are just strange
[14:34:05] <Tpenta> i vaguely recall seeing one of those but it's damned hard to pick which number is on top
[14:34:17] <timeless> actually, they're not bad
[14:34:35] <timeless> wait, which? d8 or d100? (heck, d4 is amusing when it comes to 'top' )
[14:35:47] <richlowe> hrm, 2006/573 timed out yesterday.
[14:37:32] <Tpenta> ummm no, 2006/573 is Human-readable number library routine" and it times out on Oct 24
[14:38:39] *** karrotx has quit IRC
[14:38:47] <richlowe> kshrc times out the 24th.
[14:38:48] <sickness> hi all
[14:38:55] <richlowe> the subject of of 573 states the 17th
[14:39:07] <Tpenta> but if it converges it could be approved tomorrow
[14:39:15] <Tpenta> it must have been extended
[14:39:30] <Tpenta> Status:         waiting fast-track 10/24/2006
[14:39:52] <richlowe> Tpenta: also, do you know why 569 was run in private?
[14:39:58] <richlowe> given it's of such *direct* relevence to us?
[14:40:29] <Tpenta> ye s i do
[14:40:37] <richlowe> sure, pretty much all the case material is public, I'm reading it now.
[14:40:48] <richlowe> but it annoys me that it's impossible to contribute to something that's of direct concern.
[14:41:27] <Tpenta> the case is about changing the mechanics of how things work now to enable things to work in an open manner. there are (I believe) parts of stuff that is still very internal that need to be sorted as well
[14:42:17] <richlowe> then perhaps not posting the mail archive would be a good idea?
[14:42:19] <richlowe> :)
[14:42:27] * timeless chuckles
[14:42:50] <Tpenta> for example, does the open community really need to be a part of how we run the internal scripting stuff for the mail archives?
[14:43:00] <Tpenta> to that I have no answer
[14:43:31] <Tpenta> I think the questino about why it was run internally was asked very early on
[14:44:00] <richlowe> it was, but I haven't seen it answered yet.
[14:44:40] <timeless> what controls whether snv will give me a panic message/coredump/reboot/not reboot when things go bad?
[14:44:43] <Tpenta> I thouht it had been answered, but I'm just about to hit the sack and really dont wantg to go searching through my email
[14:44:58] <twincest> time: dumpadm for cores and -k for entering kmdb
[14:45:19] <timeless> hrm, it says /var/crash/unknown
[14:45:33] <timeless> savecore enabled
[14:45:45] <timeless> dump device: none (dumps disabled)
[14:45:55] * timeless goes to read the man page
[14:46:59] <timeless> ok, so w/o the dumpdevice this isn't doing anything, cute :)
[14:47:20] <richlowe> Tpenta: if it was, I'll find it, I'm reading the whole thing.
[14:48:12] *** sparkleytone has quit IRC
[14:48:51] *** sparkleytone has joined #opensolaris
[14:50:21] *** sparkleytone has quit IRC
[14:50:54] *** sparkleytone has joined #opensolaris
[14:50:56] <richlowe> Hm, maybe it's a good thing this isn't open, I think I'd be yelling by about now.
[14:54:46] *** laca has joined #opensolaris
[14:56:42] <SymmWork> question, does anyone know what the IGMP v3 support in solaris is like at the moment?
[14:56:48] <SymmWork> (ipv6 multicast stuff)
[15:11:00] <timeless> can i not use zfs for a dump volume?
[15:11:09] <richlowe> No.
[15:11:15] <richlowe> You can swap to it, but not dump to it.
[15:11:21] * timeless sighs
[15:11:45] <timeless> can i ask the system to dump me into the kernel debugger when it panics?
[15:11:51] <twincest> boot with -k
[15:12:07] <timeless> can i do it w/o rebooting? :)
[15:12:36] <richlowe> Yes.
[15:12:49] <timeless> mdb -K ?
[15:12:52] <richlowe> on the console, (and I do *mean* the console, without X running, etc), run mdb -K
[15:12:57] <richlowe> then :c to continue.
[15:13:06] <richlowe> kmdb will remain modloaded, and re-appear on panic.
[15:13:12] <timeless> if i ctrl-alt-f1 to teh console, will i lose X11? :)
[15:13:17] * timeless had that w/ some *nix
[15:13:20] <twincest> you can't ctrl-alt-f1 to the console
[15:13:25] <richlowe> if youc ctrl-alt-f1 nothing will happen.
[15:13:28] <richlowe> vconsole isn't integrated yet :)
[15:13:33] <timeless> wfm
[15:13:36] * timeless logs out
[15:13:48] <twincest> is there any eta on that?
[15:14:36] <timeless> actually
[15:14:37] * timeless ponders
[15:14:45] <timeless> i shouldn't need x11 to do this testing
[15:15:17] <timeless> ok
[15:15:18] <timeless> it panic'd
[15:15:22] <timeless> :)
[15:15:38] <timeless> "now what" :)
[15:16:02] <timeless> i don't suppose i can write files to a usb file system from mdb :)
[15:16:03] <richlowe> now you wish you could read the mdb manual at the same time :)
[15:16:07] <richlowe> nope.
[15:16:20] <timeless> yeah, well, um, it's dead jim :)
[15:16:23] <richlowe> ::msgbuf will give you the message buffer.
[15:16:30] <richlowe> ::cpuinfo -v will give basic info about the CPU
[15:16:49] <richlowe> $C will give you a stack trace, as should *panic_thread::findstack
[15:16:59] <richlowe> (there's cases where they won't be *the same* stack, however)
[15:17:14] <timeless> ok, bad tryp type=e #pf Page Fault, ip due to a null pointer dereference
[15:17:14] <richlowe> ::help is basic help, ::dcmds is a list of commands
[15:17:22] <richlowe> ::help <command name> is more help on a command. :)
[15:17:23] <timeless> trap even
[15:18:05] <richlowe> if you have the ability to temporarily add something to dump onto, it maybe easier to let it dump via a temporarily solution...
[15:18:14] <timeless> i guess it's ip:ire_local_same_illl_group+1c()
[15:18:20] <richlowe> (I *think* you can swap/dump onto an NFS mounted file...)
[15:18:25] <timeless> i presume that a dvd writer doesn't count
[15:18:38] <richlowe> though probably not if you're panicing in IP ;)
[15:18:39] <timeless> hrm, i /might/ be able to ask someone to offer me an nfs mount
[15:18:43] <timeless> ...
[15:19:05] <richlowe> back up the USB drive and dump onto that? :)
[15:19:18] * timeless checks the usb drive
[15:19:24] <timeless> 64m from microsoft
[15:19:34] <timeless> minidumps from windows would fit there :)
[15:19:53] <timeless> 64k should be good enough for everyone. s/k/m/
[15:20:14] <timeless> lemme ask aroudn
[15:20:24] *** MattMan has quit IRC
[15:20:27] <timeless> i have 4gb of ram, i presume i need a device w/ 4+gb ?
[15:20:52] <richlowe> it only dumps the kernel pages by default.
[15:20:57] <richlowe> at worst, yes.
[15:21:31] <richlowe> the last couple I threw myself at were 108M-ish from a machine with 2G of physical memory.
[15:21:46] <richlowe> though it didn't run long before I induced the panic(s)
[15:22:16] <richlowe> I think the dump code compresses them on the dump device, too, but I don't know how well.
[15:24:28] <timeless> ok, i have 2 4g SD things
[15:24:35] <timeless> would one of those work in a usb reader?
[15:26:54] *** Vanuatoo has quit IRC
[15:27:33] <richlowe> I'm still trying to work out how you don't have *anything* suitable as a dump device. :)
[15:27:38] <richlowe> swapping to a zvol and nothing else?
[15:28:29] <timeless> i have 4gb of ram!
[15:28:48] <timeless> and i have a 5gb zvol that i hadn't needed to swap -a
[15:28:56] * timeless kinda forgot to do that a while ago
[15:29:10] <timeless> anyway
[15:29:19] <timeless> i now have a 4gb sdcard in my sdreader (usb)
[15:30:02] <timeless> ok, kernel /path -k ?
[15:30:54] *** peteh has quit IRC
[15:31:07] *** nachox has joined #opensolaris
[15:31:35] <timeless> ok
[15:31:49] *** qbit has left #opensolaris
[15:31:54] <timeless> will i be able to convince the usb reader to enable w/o using x11?
[15:32:39] *** logic has quit IRC
[15:32:40] *** logic_ has joined #opensolaris
[15:33:16] * timeless sighs
[15:33:27] <timeless> /rmdisk/unnamed_rmdisk is not valid for swapping
[15:34:01] <timeless> /vol/dev/dsk/c2t0d0 is not valid for swapping
[15:34:58] *** jamesd_ has joined #opensolaris
[15:35:07] <timeless> ok, man swap claims i can use nfs
[15:35:15] <timeless> but that seems like it won't work
[15:35:28] <timeless> i think it's paper/pencil time
[15:39:47] * timeless ponders
[15:39:50] <timeless> df .
[15:40:14] <timeless> /rmdisk/unnamed_rmdisk(/vol/dev/c2t0d0/unnamed_rmdisk:c): 8089536 blocks      -1 files
[15:40:17] <timeless> ^ -1 ? :)
[15:42:07] <richlowe> impressive.
[15:42:10] *** svoboda has quit IRC
[15:42:34] *** jamesd has quit IRC
[15:43:39] <timeless> i presume mkfile 3950M swapable
[15:43:47] <timeless> will work and give me a prompt eventually
[15:46:37] *** mrdeviant has joined #opensolaris
[15:47:17] *** calLNCH is now known as calumb
[15:59:00] *** besonen_mobile has quit IRC
[16:01:08] *** yydzero has quit IRC
[16:13:46] *** glagasse has joined #opensolaris
[16:15:47] <timeless> ok
[16:16:04] <timeless> how do i convince this unnamed_rmdisk thing to obe something that solaris will swap to?
[16:17:08] <jamesd_> you should be able to treat it like any other disks... label it and make your slices and format and enable swap
[16:21:04] *** Mazon is now known as mazon
[16:21:51] *** calum_ has joined #opensolaris
[16:21:54] *** svoboda has joined #opensolaris
[16:24:01] <timeless> labelit /vol/rdsk/unnamed_rmdisk
[16:24:10] <timeless> labelit: bad super block magic number
[16:25:25] *** calumb has quit IRC
[16:29:43] <lasseoe> timeless, if you turn off vold you can use 'format' to slice it and use it normally like any otehr disk
[16:29:51] <bengtf> timeless: must format first if you want it as a ufs volume , but you need to disconnect it from the volumemanager or try to make a sufficiently large file on it
[16:30:19] <bengtf> swap might not be able to swap to a file on vold either
[16:31:15] <timeless> o hope there's a safe+easy way to turn off vold
[16:31:24] <bengtf> but then I have had problems with formatting on late versions, pristine pata disks refuses to be labeled
[16:31:41] <bengtf> or driver doesnt reload data
[16:32:12] *** loke has joined #opensolaris
[16:33:03] *** nwf has joined #opensolaris
[16:33:11] *** besonen_mobile has joined #opensolaris
[16:33:14] *** Vanuatoo has joined #opensolaris
[16:35:00] * timeless frowns
[16:35:07] * timeless is so very confused
[16:35:54] <timeless> can i really do this?
[16:36:00] <timeless> all i want to do is swap to an sdcard
[16:36:05] <timeless> or really panic to it
[16:38:42] *** salamanders has joined #opensolaris
[16:39:52] <bengtf> yeah, shut down vold shouldnt be a problem unless you plan to attach usb or other stuff while panicing ;)
[16:40:58] <timeless> i think i need specific steps
[16:41:10] <timeless> actualy
[16:41:21] <timeless> let's try this the other way
[16:41:27] <timeless> i have .8G free on / (UFS)
[16:41:49] <timeless> can i create a dump file *there* and ask the os to please not spend more than .8G on that?
[16:42:26] <bengtf> you just need a swap file that is big enough, 500 meg should possibly enoguh unless you load extremly much at startup
[16:42:35] <timeless> i don't
[16:42:41] <timeless> at least, not intentionally
[16:42:52] <timeless> hrm
[16:42:54] <timeless> something's wrong
[16:43:30] <timeless> ok, mkfile or ufs was sillly
[16:43:43] <timeless> i asked it for an 800G file, and it gave me a note no space left on device
[16:43:52] <timeless> and ls says it's 800g :)
[16:43:56] *** jcea has joined #opensolaris
[16:47:16] *** MattMan has joined #opensolaris
[16:48:49] <Stric> timeless: and du(1) says?
[16:49:17] <timeless> i killed it because i want some space in my root volume, so now i get to wait for a 512m file :(
[16:52:31] *** mdj has joined #opensolaris
[16:54:20] *** drifter747 has joined #opensolaris
[16:56:32] *** eboutilier has joined #opensolaris
[16:56:36] *** laca is now known as lacaMtg
[16:58:14] * elektronkind grabs his rake, looks around for some muck.
[17:02:50] <timeless> elektronkind: visit me, i have lots of muck
[17:09:59] *** bank has joined #opensolaris
[17:12:18] <bank> hi all :)
[17:12:34] *** Auralis has quit IRC
[17:12:40] <jamesd_> hi
[17:12:52] *** Auralis has joined #opensolaris
[17:15:51] *** schily__ has joined #opensolaris
[17:15:54] <timeless> ok, how do i land into the debugger, is there a sysreq thing?
[17:19:04] *** trs81 has joined #opensolaris
[17:19:36] <quasi> mdb -K or something like that
[17:19:48] <timeless> i don't have a console
[17:19:54] <timeless> because the one app using my console won't talk to me :)
[17:19:59] <richlowe> you mean with kmdb loaded already?
[17:20:03] <richlowe> (and sparc or x86?)
[17:20:04] <timeless> yeah
[17:20:07] <timeless> x86
[17:20:10] <richlowe> Stop-A on sparc, F1-A on x86
[17:20:34] * timeless wonders if this system will still boot
[17:21:03] <timeless> svc.startd[7] doesn't seem to like svc:/system/identity:node :(
[17:21:10] *** mazon is now known as Mazon
[17:24:49] *** sommerfeld has joined #opensolaris
[17:24:49] *** ChanServ sets mode: +o sommerfeld
[17:27:03] * timeless really needs an online mdb reference
[17:27:19] * timeless tries http://www.cs.duke.edu/cgi-bin/man.cgi?section=1&topic=kmdb
[17:27:22] *** schily_ has quit IRC
[17:28:08] <richlowe> docs.sun.com, search "Modular Debugger Guide"
[17:30:05] <quasi> I think there's one on solarisinternals.com as well
[17:33:38] *** glagasse has quit IRC
[17:35:17] *** jcea has left #opensolaris
[17:40:01] <dclarke> g'day
[17:41:01] *** solarisjon has joined #opensolaris
[17:41:16] *** nwf has quit IRC
[17:42:25] <jteo> dclarke: g'day.
[17:42:53] *** lacaMtg is now known as laca
[17:43:10] <dclarke> g'day Mr. Peter
[17:44:43] * timeless wonders about ufs`ufs_delete_adjust_stats
[17:45:29] <timeless> so, if all i have is mdb, can i use dtrace :)? :)
[17:47:59] <jteo> timeless: orthogonal.
[17:48:34] *** calum_ has quit IRC
[17:49:21] <timeless> jteo: i think my system is hung doing something, i'd like to know if i'm right
[17:49:47] <timeless> if my system was alive, i'd try to ask dtrace to tell me (after i figured out how to do that, which would take a few days)
[17:50:06] <dclarke> you "think" ?  your system is hung ?
[17:50:17] <dclarke> generally its pretty clear when it happens
[17:50:26] <dclarke> do you have a console prompt of any type ?
[17:50:36] <timeless> ok, it didn't finish booting, it almost certainly was planning to switch to maintenance mode
[17:50:45] <timeless> but it never got anywhere close to giving me a login prompt
[17:51:03] <dclarke> uh huh .. okay .. could be something jammed up in the boot process
[17:51:07] <dclarke> smf will tell all
[17:51:12] <dclarke> if we can get a prompt
[17:51:13] <timeless> smf?
[17:51:22] <dclarke> service manifest facility
[17:51:22] <timeless> yeah well, um... i have mdb and that's it :)
[17:51:27] <dclarke> oh ..
[17:51:42] <timeless> i mean, it's not 100% dead, since f1-a gets me mdb
[17:51:43] <dclarke> can you get back to a console ?  I mean OBP ?
[17:51:46] <timeless> and i can see the process list
[17:51:51] <timeless> obp?
[17:51:55] <dclarke> open boot prompt
[17:52:02] <timeless> ?
[17:52:07] <dclarke> I guess this is not Sparc eh ?
[17:52:13] <timeless> i can always power cycle the box, this is x86
[17:52:14] <dclarke> you're on x86/AMD64 ?
[17:52:22] <dclarke> okay .. so no boot firmware
[17:52:23] *** stevel has joined #opensolaris
[17:52:24] *** ChanServ sets mode: +o stevel
[17:52:25] <timeless> technically amd64 i suppose
[17:52:26] <dclarke> dumb x86
[17:52:41] <dclarke> I personally love having an OBP
[17:52:55] <timeless> macsbug has been fun in the past
[17:53:03] <timeless> although i kinda liked bdb better
[17:53:05] <dclarke> okay .. so at best you can get to grub and then from grub
[17:53:18] <timeless> yeah, i can get to grub if i need to
[17:53:19] <dclarke> boot the kernel in single user mode ..
[17:53:22] <dclarke> or the debugger
[17:53:24] <timeless> and i have the dvd so i could reinstall
[17:53:27] <icon> boot -s
[17:53:36] <dclarke> well ... boot -sv
[17:53:45] <icon> true enough ;)
[17:53:51] <dclarke> or .. boot  the kernel .. specifically
[17:53:52] <icon> morning dclarke
[17:53:57] <dclarke> morning
[17:54:01] * dclarke checks time
[17:54:07] <dclarke> yep .. still morning
[17:54:13] <icon> est ?
[17:54:23] <timeless> ok, so it looks like zfs is in
[17:54:30] <timeless> http://cvs.opensolaris.org/source/xref/on/usr/src/uts/common/fs/ufs/ufs_thread.c#630
[17:54:36] <timeless> which isn't particularly interesting
[17:54:56] <dclarke> I'm a tad busy fighting with some fibre arrays
[17:55:01] <dclarke> or I'd go look at that
[17:55:13] <timeless> you're probably fighting something more useful
[17:55:16] <richlowe> why would zfs have called into ufs?
[17:55:22] <richlowe> or was that a typo?
[17:55:56] <timeless> it's not a typo
[17:56:12] *** mkhl has joined #opensolaris
[17:56:16] <timeless> _sys_sysenter_post_swapgs
[17:56:20] <timeless> statvfs32
[17:56:25] <timeless> cstatvfs32
[17:56:31] <timeless> fsop_statfs
[17:56:36] <timeless> ufs`ufs_statvfs
[17:56:38] <timeless> ...
[17:56:46] <timeless> this is using:
[17:56:58] <timeless> ::pgrep zfs | ::walk thread | ::findstack
[17:57:58] <sommerfeld> timeless: what process was this?
[17:58:04] *** solarisjon has left #opensolaris
[17:58:16] <timeless> there are only 12 lines in ::ptree <sched <fsflush, pageout, init <snmpd, netstrategy x3, svc.configd, svc.startd <devices-local <zfs [i am here]>>>
[17:58:21] <timeless> >
[17:58:22] <richlowe> oh, zfs(1)
[17:59:33] *** calumb has joined #opensolaris
[17:59:52] <timeless> so what's this delete thread that ufs_delete_adjust_stats is talking about :)
[18:00:48] *** Cyrille has quit IRC
[18:00:50] *** Dar is now known as Dar_HOME
[18:00:55] <sommerfeld> when the refcount of a vnode goes to zero, the vnode gets queued on a delete queue, and the delete thread actually frees the space it occupied
[18:01:02] <sommerfeld> (for ufs, at least.)
[18:01:08] <sommerfeld> I think zfs does something similar
[18:01:16] <timeless> is there some convenient way for me to chase that thread? :)
[18:01:42] <jamesd_> dtrace
[18:01:57] <timeless> can i start dtrace from mdb? :) - i think we were here earlier
[18:02:49] <sommerfeld> i missed the start of this -- how did you end up shaving this yak?
[18:03:11] <timeless> the start is that i managed to use a zone to panic the os using ping
[18:03:13] <alanc> sure - ^Z mdb and then type dtrace at the prompt
[18:03:16] <timeless> that panic is in ip somewhere
[18:03:31] <richlowe> Hm.
[18:03:34] <timeless> then i tried to find a place for a dump thing
[18:03:46] <timeless> and the only place available was /, because i tried everything else
[18:03:55] *** slowhog has joined #opensolaris
[18:04:06] <timeless> well, richlowe said trying to use nfs for an ip panic was a bad idea, so i didn't bother trying that
[18:04:24] <timeless> then i made the mistake of trying to make an 800g file on a volume w/ <1g free
[18:04:30] <timeless> instead of an 800m file
[18:04:54] <timeless> the file was created using all available space and the os told me (root) that i was out of space
[18:05:08] <timeless> i hastilly used rm to kill that and tried to use mkfile again to make a 512mb file
[18:05:16] <timeless> this didn't work, so i rebooted
[18:05:38] <timeless> this time, it didn't boot very far, but i finally knew how to use f1-a
[18:05:43] <richlowe> ... you told it to dump onto /
[18:05:44] <richlowe> ?
[18:05:52] <timeless> i didn't get that far
[18:06:03] <timeless> i was trying to make a file /dumpfile (512mb)
[18:06:08] <jteo> why don't you just pull 2 sticks of RAM out of the box.
[18:06:10] <timeless> you're supposed to be able to ask it to dump to files
[18:06:19] <timeless> how would that help?
[18:06:20] <jteo> then your corefile will be small enough to fit in swap
[18:06:25] <timeless> i don't have swap!
[18:06:34] <timeless> i have 5gb of swap in zfs
[18:06:37] <jteo> no, you don't have enough space for swap.
[18:06:40] <timeless> which isn't good enough for dumps
[18:06:49] <jteo> and you didn't configure swap because?
[18:06:53] <timeless> and i have 1gb memory modules
[18:07:13] <timeless> because i don't expect to need it w/ 4gb of ram, and planned to use zfs for swap if necessary
[18:07:16] <richlowe> jteo: swapping to a zvol works, dumping to a zvol doesn't.
[18:07:22] *** damienc has quit IRC
[18:07:27] <jteo> richlowe: yup.
[18:07:33] <timeless> there was no expectation that i was going to need to deal w/ panics
[18:07:47] *** damienc has joined #opensolaris
[18:07:48] <timeless> and certainly i didn't read the fine print about swap v. dump re zvol
[18:07:58] <sommerfeld> murphy's law at work: if you do not provide space for a crash dump, you will need it. :-)
[18:08:03] <jteo> timeless: i'll skip the pointless whys. ;)
[18:08:11] <timeless> sommmerfeld: yeah well
[18:08:12] <jengelh> why?
[18:08:21] <timeless> jengelh: i hope that's to jteo ;-)
[18:08:26] *** bengtf has quit IRC
[18:08:26] <sommerfeld> so, this is amd64 with zfs, so you have newboot at work.
[18:08:31] <jengelh> heh yeah
[18:08:51] <sommerfeld> have you been able to successfully do a failsafe boot?
[18:08:52] <timeless> the reason i'm using zfs in the first place is that i had already run out of disk space on this puny 80gb drive fwiw
[18:09:00] <timeless> i haven't tried that yet
[18:09:19] <sommerfeld> ok.  what I would recommend would be to do a failsafe boot, which will get you running in a ramdisk /
[18:09:21] <timeless> i'd expect the failsafe to be just as pissed about the ufs volume as the normal boot
[18:09:42] <sommerfeld> i would then run fsck on the root filesystem
[18:10:28] <timeless> ok, which process has this ufs delete thing?
[18:10:41] <sommerfeld> it's a kernel thread which gets started when you mount the filesystem.
[18:10:54] <timeless> hrm, so it's not in ::ptree
[18:10:58] <sommerfeld> right.
[18:11:02] <timeless> 1 ::walk thread || ::findstack
[18:11:03] <timeless> ?
[18:11:11] <sommerfeld> yep
[18:11:26] <richlowe> nope.
[18:11:29] <timeless> not 1
[18:11:31] <richlowe> one |, not two.
[18:11:37] <richlowe> and not 1, because that needs to be a proc address.
[18:11:52] <richlowe> if 1 is intended to be a pid 1::pid2proc | ::walk thread | ::findstack
[18:11:59] <timeless> thankfully i'm not retyping the tail of this expression so i can get it wrong here :)
[18:12:14] <sommerfeld> oh, whoops, leave out the 1
[18:13:05] <timeless> ok, i'm going to do ::threadlist
[18:13:09] <timeless> that seems like a better idea
[18:13:30] <timeless> ok, i see ufs`ufs_thread_delete()
[18:13:33] <timeless> i presume that's what i want
[18:13:41] <sommerfeld> so, chances are that the delete thread is busy deleting that 800g file
[18:14:52] *** salamanders has quit IRC
[18:15:08] <timeless> thread_start/ufs`ufs_thread_delete/ufs`ufs_delete/ufs`_ufs_trans_itrunc/ufs`ufs_trans_trunc_resv/ufs`ufs_log_amt/_resume_from_idle
[18:15:35] <timeless> ok, well, at this point, if i had a working computer, i think i'd be ready to debug my real panic :)
[18:15:53] <sommerfeld> yup, it's trying to delete that file
[18:16:39] <timeless> ok, i think my next question is "what cpu is that thread on"
[18:16:43] <elektronkind> try booting into failsafe and mounting that fs without logging on
[18:17:26] *** slowhog has quit IRC
[18:18:21] <bank> :)
[18:19:12] *** koolniczka has joined #opensolaris
[18:20:55] *** loke has quit IRC
[18:23:43] <timeless> ok, so, cpu2 (0 based) is assigned to this task
[18:24:20] *** astinus has joined #opensolaris
[18:24:42] *** bank has quit IRC
[18:24:59] <sommerfeld> the fast way out here may well be to find the inode number, use clri, and then fsck to recover the lost disk space.
[18:25:32] <timeless> sounds like a good plan
[18:25:53] <timeless> i'm learning about what RNRN yes and SWITCH t-68206 (::cpuinfo) mean atm
[18:26:30] <timeless> and learning that -v is my friend :)
[18:26:41] <timeless> or not
[18:26:47] * timeless still has no clue what rnrn means
[18:28:27] <timeless> ok, i can't find anyone who has a RNRN yes on the web :)
[18:29:57] <richlowe> rnrn is runrun I think, preemptable, or preempted (or about to be preempted?)
[18:30:59] <timeless> http://blogs.sun.com/esaxe/date/200506 seems to try to explain it
[18:31:28] <timeless> i think i probably want to read that thing top to bottom
[18:32:27] *** damienc has quit IRC
[18:34:50] <richlowe> yeah, RNRN and KRNRN are cpU_runrun and cpu_kprunrun, which trigger preemption.
[18:35:42] <timeless> does that mean that this thread /wants/ to be preempted?
[18:36:03] <richlowe> it's a flag on the CPU, not the thread.
[18:36:27] <richlowe> it means a higher priority thread has been placed on a dispq, and that the CPU should preempt what's currently running on it in favour of that thread, I *think*
[18:36:32] <movement> timeless: it means that something has set the flag and will cause a pre-emption
[18:36:45] *** MattMan has quit IRC
[18:37:04] <PerterB> I think you're right richlowe (and the kprunrun is only used for RT class threads if I recall)
[18:37:25] <timeless> if that's the case, should i see a sign of the "something" somewhere?
[18:37:39] <timeless> because there aren't many things floating around in ::cpuinfo -v
[18:38:55] <timeless> oh wow, in this one example he had krnrn at yes, i'll be able to learn something about it there i expect :)
[18:41:06] *** slowhog has joined #opensolaris
[18:46:04] * timeless reads about ::cycinfo -v
[18:48:30] *** eboutilier has quit IRC
[18:49:02] <timeless> ok, so that was an interesting asside
[18:49:17] <timeless> how do i find the inode number? :)
[18:49:51] *** sporq_ is now known as Sporq
[18:51:23] * timeless frowns
[18:51:39] <timeless> the pdf from solarisinternals.com references ::context which kmdb doesn't know
[18:52:28] <richlowe> the kmdb_kvm target doesn't, the kvm target does, iirc.
[18:52:39] <richlowe> so you can't use it within boot -k, or mdb -K, but can with dumps or mdb -k
[18:53:13] <timeless> ok
[18:54:18] <movement> it's pretty difficult to use alas
[18:54:36] <movement> once upon a time I had a workspace that made it use libproc, but that's long since bitrotted
[18:54:49] <timeless> libproc?
[18:54:51] <movement> but ::gcore from a kernel dump was pretty nifty
[18:55:02] <richlowe> movement: how long is 'long'?
[18:55:17] <richlowe> (long enough even merging it up to openable would suck?)
[18:56:01] <movement> last touched february 2005
[18:56:29] <timeless> that doesn't sound too bad :)
[18:56:58] <movement> in fact I don't even know where it is...
[18:57:04] *** dunc has quit IRC
[18:57:13] <timeless> ok, how do i change my debugger so it's actually debugging cpu2 where this process is?
[18:57:15] <movement> richlowe: if you're interested...
[18:57:35] <richlowe> timeless: what information exactly are you looking for?
[18:57:40] <richlowe> because I can't think of an answer better than "You don't"
[18:57:49] *** dunc has joined #opensolaris
[18:58:11] <timeless> i just want to be able to see what ufs`ufs_log_amt is doing
[18:58:12] <timeless> i suppose
[18:58:13] <richlowe> movement: interested does not necessarily mean able, that's why I was asking about only merging it up to June 2005, etc.
[18:58:28] <movement> richlowe: it's more a lot of work than hard work
[18:58:29] <richlowe> timeless: <that thread address>::findstack -v
[18:58:34] <movement> if you swim
[18:58:47] <timeless> [ fffffe80022d8d0 _resume_from_idle+0xf8() ]
[18:59:16] <timeless> fffffe80022da30 ufs`ufs_log_amt+0x18a(ffffffff82dca018, c77115cf12, a4d1e, 1)
[18:59:21] <richlowe> movement: I already have a collection of "things I really want to do to mdb", somewhat languishing as my list of things to do rearranges itself.
[18:59:22] * timeless goes to read ufs_log_amt
[18:59:34] <richlowe> starting off with doing the munges thing properly, finally, so it can actually exist.
[18:59:38] <richlowe> and a couple of things surrounding it.
[19:00:30] <movement> I realised my wx backup was on a home dir that got killed :/
[19:01:22] <movement> OK I found it.
[19:01:48] <timeless> http://cvs.opensolaris.org/source/xref/on/usr/src/uts/common/sys/fs/ufs_inode.h #238
[19:02:05] <timeless> wwo
[19:02:05] <movement> richlowe: let me try a merge and see what happens.
[19:02:10] <timeless> ::print inode_t worked
[19:04:03] <timeless> ok, inode->i_ic.ic_lsize=0xc7712aa000
[19:05:10] <timeless> that would be kinda like 800g i suppose
[19:05:24] <timeless> yep, very close :)
[19:06:24] <timeless> sommerfeld/movement: can you tell me which field in http://cvs.opensolaris.org/source/xref/on/usr/src/uts/common/sys/fs/ufs_inode.h inode_t is the one i want to have?
[19:06:34] <timeless> i_number?
[19:06:45] <timeless> (definitely not)
[19:09:09] * timeless reads http://cvs.opensolaris.org/source/xref/on/usr/src/cmd/fs.d/ufs/clri/clri.c
[19:09:48] *** jcea has joined #opensolaris
[19:11:43] *** bondolo has joined #opensolaris
[19:13:30] *** bougie has joined #opensolaris
[19:16:58] <movement> timeless: what field /do/ you want?
[19:17:33] <timeless> <sommerfeld> the fast way out here may well be to find the inode number, use
[19:17:33] <timeless> +clri, and then fsck to recover the lost disk space.
[19:17:46] <timeless> node->i_number = 0x54cd
[19:17:58] <timeless> i'm just a bit surprised that it'd be that "small"
[19:18:18] <timeless> keep in mind, i'm a newbie, i know nothing about ufs
[19:18:26] <timeless> i suppose inodes could be random
[19:18:27] *** dunc has quit IRC
[19:18:46] <sommerfeld> how large is your root partition?
[19:18:51] <timeless> 4gb
[19:19:10] <timeless> fairly (intentionally) small solaris config
[19:19:15] *** stevel changes topic to "opensolaris.org temporarily down | Latest releases: SXCR: 50, ON build: 50, ON nightly: 20061016"
[19:19:17] <timeless> maybe <22,000 files is right
[19:19:36] * timeless hopes cvs.opensolaris.org is still up :)
[19:19:38] * movement starts the mother of all bringovers
[19:19:57] <richlowe> if it's going to suck that hard, don't put yourself through it.
[19:20:35] <movement> it's just slow, that's all
[19:20:41] <movement> I'll see how bad the conflicts are...
[19:22:14] <gdamore> hi *
[19:22:25] <gdamore> so SXCR 50 is available now.  woohoo.  i need to take a look.
[19:22:49] * timeless wonders how gdamore plans to do that
[19:22:51] <quasi> gdamore: since friday
[19:23:01] <gdamore> as i see now.... os.org is down.  nice.
[19:23:28] *** px3^logging has joined #opensolaris
[19:23:54] <gdamore> so why is os.org down?  this is strangely reminiscint of ftp.netbsd.org outages lately.
[19:24:19] <timeless> sommerfeld: so does this number seem ok?
[19:24:44] <sommerfeld> it's not implausible
[19:25:07] <px3^logging> hi
[19:25:08] <sommerfeld> inode number usage is somewhat sparse
[19:25:29] <timeless> ok, so reboot into which specific mode to kill this thing?
[19:26:00] *** px3^logging is now known as px3
[19:26:07] <sommerfeld> try the "failsafe boot" option from the grub menu
[19:26:12] <timeless> ok
[19:26:27] <sommerfeld> you'll be running with your root filesystem from a ramdisk
[19:26:34] <px3> has anyone experience with percormance differences between TS and FSS ?
[19:26:49] * timeless nods
[19:27:01] *** musasabi has joined #opensolaris
[19:27:34] *** stevel changes topic to "Latest releases: SXCR: 50, ON build: 50, ON nightly: 20061016"
[19:27:57] * timeless wonders how to calculate what +0x18a is in that function (w.o building the kernel)
[19:28:02] <timeless> hello musabi
[19:28:12] <musasabi> hello
[19:28:54] <timeless> richlowe: the real question i had in mind is: can i step through this stack frame and see what it's doing?
[19:29:02] <musasabi> Is there an easy way to get (Open)Solaris installed and booting mostly from a network? The machine has CDROM + can boot from USB sticks, and I'd like to have it booting from USB stick and then mounting everything from a non-solaris NFS server.
[19:29:15] <timeless> oh GOD
[19:29:16] <timeless> nm
[19:29:19] <timeless> i know what it's doing
[19:29:41] <timeless> it's spinning in a for () loop in ufs`ufs_trans_trunc_resv
[19:29:47] <timeless> line 774
[19:30:17] <timeless> at least, i hope i'm right :)
[19:30:43] <px3> so, noones here with performance-experiences ?
[19:31:24] <richlowe> Hm, I know nearly nothing about UFS it seems.
[19:31:44] <timeless> http://cvs.opensolaris.org/source/xref/on/usr/src/uts/common/fs/ufs/ufs_trans.c#774
[19:32:17] <timeless> richlowe: anyway, if you could walk me through walking this thread through a couple of iterations, i'd be able to prove it to myself
[19:32:33] <timeless> i guess ::step
[19:32:52] <richlowe> If that's on the CPU, and likely to stay there, probably.
[19:33:13] <timeless> it's the only thing on cpu2
[19:33:35] <timeless> most processes are sleeping waiting for ufs to finish cleaning up :)
[19:34:26] <timeless> step - single-step target to next instruction
[19:34:41] <timeless> what's "target" and how do i make it cpu2/thread there? :(
[19:34:46] <richlowe> finding the top of that loop in ::dis, and sticking a breakpoint there maybe reasonably indicative, too.
[19:34:47] <timeless> (that's why i asked how to select the cpu)
[19:35:02] <richlowe> the target in your case is the kernel in general.
[19:35:05] <timeless> ok, i think i can do that
[19:35:42] * timeless tries to figure out how [ addr ] ::dis ... [addr]
[19:37:27] <timeless> ok, ufs`ufs... ::dis is good
[19:37:40] <timeless> i don't suppose i can get the debugger to use 50 lines instead of 24? :)
[19:39:40] <timeless> ufs`ufs_log_amt:b
[19:39:41] <timeless> :)
[19:39:56] <timeless> [2]> yippee, cpu change :)
[19:41:16] * timeless pulls out an asm ref
[19:41:24] *** px3 has quit IRC
[19:41:37] *** mdj has quit IRC
[19:42:08] *** calumb has quit IRC
[19:45:55] <icon> anyone tried a dual core opteron yet?
[19:46:57] * timeless tries to remember how to ask mdb what this cpu is
[19:47:21] <icon> thinking of picking up a 180 to replace my 150
[19:47:58] * timeless isn't sure how to distinguish dualcore from hyperthreading here
[19:48:21] <icon> well, amd chips never supported htt :)
[19:48:39] <timeless> oh, this is a xeon box says the logo :)
[19:49:01] <icon> im lazy, i usually just grep dmesg :)
[19:49:23] <timeless> i don't remember if i can get to dmesg from the debugger
[19:49:25] <icon> hrmm looks like a 180 is the best i can do with that socket 939
[19:49:29] <timeless> looking for pretty stickers is much easier
[19:49:31] <stevel> 185
[19:49:36] <icon> 185?
[19:49:38] <icon> whats the diff?
[19:49:44] <icon> (just looking on newegg atm)
[19:49:58] <stevel> socket 939 dual-core 2.6ghz
[19:50:17] <icon> ahh 200 mhz jump
[19:50:22] <icon> same l1 and l2?
[19:50:26] <stevel> 2MB L2
[19:50:33] <icon> 2MB+2MB ?
[19:50:42] <stevel> i forget the L1
[19:50:48] <timeless> richlowe: ok
[19:50:52] <icon> or is it 1 per core?
[19:50:54] <timeless> it's definitely in this loop
[19:51:05] <timeless> i'm pretty sure that being in this loop is a bad idea
[19:51:17] <richlowe> I don't suppose "continue, and wait for it to not be anymore" is an option? ;)
[19:51:24] * timeless chuckles
[19:51:30] <timeless> if i rewrite the loop a bit?
[19:51:44] <timeless> i don't think i have enough years to wait
[19:51:58] <timeless> keeping in mind that i really have *no* place to write down this data
[19:52:04] <timeless> and that i do think it's worth filing
[19:52:07] <timeless> what should i collect? :)
[19:52:26] <timeless> problem: user is very stupid and filled up / with an 800gb file
[19:52:27] <icon> ahh looks like 1MB+1MB
[19:53:12] <stevel> ah sorry, missed your msg - yeah 1+1
[19:53:13] <timeless> does anyone here have logs? i don't :(
[19:53:22] <icon> hrmm
[19:53:31] <icon> bit difference in price... i wonder why
[19:53:46] <icon> its a shame pacifica isnt supported in the 939's
[19:54:18] <timeless> http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=4794078
[19:55:25] *** sahafeez has joined #opensolaris
[19:56:28] <gdamore> what is Tamarack?
[19:56:39] <richlowe> vold/volfs replacement.
[19:56:43] <richlowe> opensolaris.org/os/project/tamarack
[19:56:49] <icon> hrmm
[19:57:24] <alanc> Tamarack is also the reason Solaris got HAL support in ON
[19:57:35] <gdamore> ah.
[19:58:09] <icon> ahh cant argue with that
[19:58:15] <icon> alanc: was HAL added to 50 or 51?
[19:58:17] <alanc> 51
[19:58:24] <alanc> as part of the Tamarack putback
[19:58:30] <icon> groovy
[19:58:40] <asyd> hal for hardware astration layer ?
[19:58:59] <alanc> http://www.opensolaris.org/os/community/on/flag-days/51-55/
[19:59:26] <alanc> asyd: yes - it's a project from freedesktop.org to provide OS-independent hardware interfaces for desktop apps
[19:59:36] *** Netwolf has quit IRC
[19:59:55] <asyd> yeah, I heard it a bit in Linx, dunno it's a freedesktop project
[20:00:56] <alanc> http://freedesktop.org/wiki/Software/hal
[20:01:15] <icon> its pretty slick to see in action
[20:01:17] <gdamore> so does cdrecord support full size 4GB dvds now?
[20:07:39] *** yippi has joined #opensolaris
[20:10:29] *** nachox has quit IRC
[20:12:26] *** slowho1 has joined #opensolaris
[20:13:17] *** slowho1 has quit IRC
[20:23:43] <timeless> ok, what's my synopsis? ufs`ufs_trans_trunc_resv calling ufs`ufs_log_amt a lot is bad? :)
[20:24:31] *** mrdeviant has quit IRC
[20:25:02] *** besonen_mobile_ has joined #opensolaris
[20:31:41] *** dunc has joined #opensolaris
[20:32:52] <timeless> ok. i've sent my bug report
[20:32:57] * timeless looks for reboot
[20:34:26] *** bengtf has joined #opensolaris
[20:35:07] * timeless sighs
[20:35:13] <timeless> ok, how do i reboot from mdb?
[20:35:38] <timeless> the docs.sun.com solaris modular debugger guide explains how to reboot from a shell
[20:35:42] <timeless> but i don't have a shell :(
[20:37:14] * timeless power cycles
[20:37:48] <timeless> um
[20:37:51] <timeless> failsafe *failed*
[20:38:49] * timeless wonders why
[20:38:58] <timeless> oh well, it worked this time w/ -k added
[20:39:27] <movement> timeless: from kmdb? $q
[20:40:03] *** Netwolf has joined #opensolaris
[20:40:04] <timeless> quiting the debugger will reboot in that case?
[20:40:05] <timeless> ok
[20:40:15] *** besonen_mobile__ has joined #opensolaris
[20:40:29] *** besonen_mobile has quit IRC
[20:41:07] <timeless> ok, now i just fsck /dev/dsk/c1d0s0 ?
[20:42:38] <timeless> heh, unref dir i=217067 owner=root mode=40755 size=512 mtime=oct 18...
[20:42:43] <timeless> that was my second attempt
[20:43:41] <timeless> hrm
[20:43:48] <timeless> 113526 files
[20:46:38] <timeless> ok
[20:48:02] *** besonen_mobile_ has quit IRC
[20:49:16] <timeless> ok, panic: entering debugger (continue to save dump)
[20:50:39] <timeless> actually, at this point, i think i know enough about mdb that i could probably do this debugging from it, but having a smaller font will be nice :)
[20:51:06] *** nwf has joined #opensolaris
[20:52:22] <timeless> um
[20:52:38] <timeless> panic[cpu1]/thread=ffffffff8ce33b20: panic dump timeout
[20:54:40] <timeless> http://bugs.opensolaris.org/bugdatabase/view_bug.do;jsessionid=334698776f65f7a39a5aeea89e4a:vCrm?bug_id=4063460
[20:54:41] <timeless> that's cute
[20:56:49] <AbeFroman> a double panic
[20:57:16] <timeless> it's not my day
[20:57:23] <timeless> this is panic #3
[20:57:40] <timeless> panic 2 was by me failing to properly setup to debug panic 1
[20:57:49] <timeless> panic 3 is because i properly setup to debug panic 1
[20:59:07] <twincest> where is the source for the main part of NCA?
[20:59:18] <AbeFroman> shuttle engine test
[20:59:42] <AbeFroman> whoops wrong channel
[21:01:35] * timeless frowns
[21:01:42] <timeless> [1]> *panic_thread::findstack
[21:01:51] <timeless> is giving me what i think is the *original* panic
[21:01:54] <timeless> and not the new one
[21:03:44] <timeless> heh
[21:03:50] <timeless> kthread_t *panic_thread;/* first thread to call panicsys() */
[21:03:57] *** alanc is now known as alanc_away
[21:04:10] *** Seawolf99 has joined #opensolaris
[21:05:23] *** jwtodd has quit IRC
[21:06:07] *** jwtodd has joined #opensolaris
[21:06:10] <Seawolf99> hi, I am trying to upgrade ON with bfu. just wondering if I have to switch run level or boot in single mode or something else before I do this
[21:06:13] *** jwtodd has left #opensolaris
[21:06:24] *** jwtodd has joined #opensolaris
[21:07:00] <timeless> ok, so i'm looking for a thread that's in deadman()
[21:07:16] *** woland_ has joined #opensolaris
[21:08:42] *** jwtodd has quit IRC
[21:09:12] <gdamore> hmm.... i'm thinking i might want to add synaptic tpad support to hid, and create a vuidsynps2 module.
[21:10:08] <Seawolf99> hi, I am trying to upgrade ON with bfu. just wondering if I have to switch run level or boot in single mode or something else before I do this
[21:10:59] <gdamore> run levels are fairly arbitrary if no one else is using the system, but i'd go single user to minimize possible interactions with other system software.
[21:11:38] <Seawolf99> thanks gdamore
[21:12:01] *** jwtodd has joined #opensolaris
[21:17:35] <timeless> if i boot to -s (single user), can i still have networking and zones?
[21:17:39] <timeless> i want to skip x11 :)
[21:17:52] <jengelh> timeless : Just remove the link in /etc/init.d
[21:18:14] <timeless> that requires me to boot :)
[21:18:25] <gdamore> pkill dtlogin works.
[21:18:39] <woland_> does smpatch take as long on opensolaris as it does on solaris?
[21:18:45] * timeless wonders if x11 is a svcs thing
[21:18:46] <gdamore> i don't know about zones, but boot -s will leave you with basic networking.
[21:19:04] <gdamore> dtlogin certainly is a svcs thing.
[21:19:14] *** phalenor1 is now known as phalenor-
[21:19:18] <gdamore> but i don't think svcs will restart it if it dies.
[21:19:19] <timeless> i don't see it
[21:19:22] <oxygene> last time I looked dtlogin is done with legacy_rc
[21:19:47] <gdamore> ah, yes, oxygene, you're correct on S10 3/05 at least.
[21:19:51] <timeless> oh
[21:19:57] <timeless> graphical-login/cde-login
[21:20:06] <timeless> i presume i can disable that and i'll be happy?
[21:20:25] <gdamore> probably.  i wonder if /usr/dt/bin/dtconfig -d still works. :-)
[21:20:40] <oxygene> gdamore: afaik it's a wrapper now
[21:20:52] <oxygene> "interface compatibility" and all
[21:20:57] <gdamore> heh.
[21:21:44] <gdamore> in other news, it looks like my "trivial" update to make kstat itnerfaces use const char * instead of char * is finally going to get putback.  the amount of redtape for even simple changes like this is amazing.
[21:22:03] *** gtc_ has joined #opensolaris
[21:23:41] <gtc_> Hi all -- basic question.  How can I get an ident response under b48 & irssi?  Some freenode servers refuse to let me connect.
[21:26:19] <delewis> gtc_: enable identd
[21:26:34] <timeless> ok, how do i tell solaris to only use 128mb of ram (from grub)?
[21:26:35] * delewis wasn't aware Freenode required identd
[21:27:09] * timeless grumbles
[21:27:10] <delewis> gdamore: sounds like a similar discussion on osol-discuss and osol-arc right now :-)
[21:27:15] <gtc_> delewis: I don't think I have identd.
[21:27:26] <timeless> panic[cpu3]/thread=ffffffff8ba14480: panic dump timeout
[21:27:27] <delewis> gtc_: should be a standard, inetd daemon
[21:27:48] *** logic__ has joined #opensolaris
[21:27:49] <gdamore> delewis: i'm not on either of those lists.  i wonder every now and then if i should subscribe to osol-discuss
[21:28:13] <delewis> gdamore: there's just an on-going rant about the ARC process and how time consuming it is.
[21:28:14] <gtc_> delewis: no identd in /sbin /opt/csw/sbin/ /usr/sbin /opt/csw/bin /usr/sbin /usr/bin /usr/dt/bin /usr/openwin/bin /usr/ccs/bin
[21:28:17] *** Tpenta has quit IRC
[21:28:48] <gtc_> delewis: !irc.freenode.org *** No identd (auth) response
[21:28:51] <timeless> looks like it takes 3 mins or so for the deadman to kill me the second time
[21:29:05] *** lulf has joined #opensolaris
[21:29:55] <timeless> delewis: any ideas?
[21:30:29] <delewis> timeless: since Grub, I'm not sure you can boot Solaris on a system with less than 256B of memory
[21:30:36] <delewis> (or would want to for that matter)
[21:30:58] <timeless> 256m is fine
[21:31:06] <timeless> just so long as it's << 4g
[21:31:11] <delewis> with Solaris 10/oldboot it was possible (then the "recommended" minimum was 256MB, but 128MB was acceptable)
[21:31:12] <oxygene> someone got it down to 60mb iirc (it's only the first stage of loading the ram disk)
[21:31:15] <gdamore> why would grub make the requirement more?
[21:31:23] <timeless> http://www.opensolaris.org/jive/thread.jspa?messageID=51374
[21:31:27] <timeless> talks about it requiring more
[21:31:29] <timeless> i don't care
[21:31:33] <oxygene> gdamore: because it needs to load a huge ramdisk with the basic drivers
[21:31:36] <timeless> i just want to limit the os to << than physical
[21:31:48] <timeless> all i'm doing is booting enough to panic the os
[21:31:55] <timeless> and that doesn't require x11 (which i disabled)
[21:32:10] <oxygene> there would be ways to fix that.. most of them requiring more work than is useful for the ">256mb" constraint
[21:33:13] <gdamore> i'm surprised, and disappointed.  solaris could still be useful on lower memory systems if you don't want to run a desktop.  e.g. as an NFS server.
[21:33:26] <timeless> can we focus on my problem? :)
[21:33:35] <timeless> there's a whole list where you can express your disappointment
[21:33:36] <delewis> gdamore: it's not impossible -- it just requires hacks not currently integrated.
[21:33:39] <oxygene> gdamore: as I said, someone managed to get the ramdisk to 60mb or so
[21:34:04] <jamesd_> timeless, absolutely not.. there is "time" for your problem next month
[21:34:09] *** Jondice has quit IRC
[21:34:30] <timeless> jamesd: ok :)
[21:36:49] <jerome__> Okay, I have a ZFS partition with 40 million files, the files are in nested directories such as: /zfs/3/7/9/26/268/image_268.txt, the throughput I get in random i/o read is really bad, how do I improve it? (I'm on 4x300GB 10K rpm scsi drives in raid-0)
[21:37:35] *** k2flag has joined #opensolaris
[21:38:15] <dclarke> let's see who is around ..
[21:38:16] <andersmo> jerome__: how many iops is "really bad" in your case?
[21:39:02] <jamesd_> jerome__, need more info, are they all sharing one scsi channel,  add more ram? more scsi channels, more drives?  could it be running in 32 bit mode?
[21:39:08] <jamesd_> hi dclarke
[21:39:11] <jerome__> it's like 150 tps
[21:39:22] <dclarke> jerome : I have similar problems here with a fibre array stack
[21:39:38] <dclarke> jamesd : bon jour .. long time no .. nothin
[21:40:09] <jerome__> I have 8GB of ram and 2 Opteron 265
[21:40:10] *** logic_ has quit IRC
[21:40:12] <jerome__> or maybe 275
[21:40:32] <jerome__> but could the overhead be the nested directory structure?
[21:40:35] <dclarke> I have a 4GB RAM V240 here
[21:40:38] <dclarke>     0.0    7.6    0.0 7782.4  0.0  1.0    0.0  131.6   0 100 c3t2d0
[21:40:39] <dclarke>     0.0    6.2    0.0 6348.8  0.0  1.0    0.0  161.3   0 100 c1t118d0
[21:40:40] <dclarke>     0.0    6.2    0.0 6365.9  0.0  1.0    0.0  160.8   0 100 c1t112d0
[21:40:44] <dclarke> iostat says
[21:40:50] <jerome__> gdamore: what's your take?
[21:41:13] <dclarke> disks are 100% busy they say
[21:41:16] <jamesd_> jerome__, not likely... what is the actual throughput you are seeing?
[21:41:29] <gdamore> i'm not really a disk/filesystem guy, so i'm not the best to ask
[21:41:32] <jerome__> I think I can read 150 files a second or so
[21:41:38] <jerome__> which seems really lame
[21:41:41] <dclarke> 150/sec you think ?
[21:41:45] <jerome__> yeah
[21:41:54] <dclarke> can you run a little test for me ?
[21:41:57] <jerome__> sure
[21:42:04] * dclarke expects jamesd to smake me
[21:42:05] <jamesd_> jerome__, but  how many megabytes per second?  is this ultra scsi?  u160?  u320?
[21:42:06] <gdamore> but it would be informative to find out if the problem is finding new files, or doing random io in the same file.
[21:42:28] <jerome__> I'm sorry I forgot to mention these files are EXACTLY 20k big EACH
[21:42:29] <dclarke> jerome__: let me get you a binary ..
[21:42:38] <dclarke> oh heck .. you're on Opteron
[21:42:43] <dclarke> I have to recompile
[21:42:44] <jerome__> gdamore absolutely
[21:42:47] <dclarke> just a moment
[21:42:48] <jerome__> I am indeed
[21:42:55] <andersmo> jerome__: atime?
[21:42:59] <jerome__> noatime
[21:43:14] <jerome__> I'm doing an open("file",O_RDONLY)
[21:43:23] <jerome__> on a noatime filesystem
[21:43:33] <dclarke> jerome__: what rev of Solaris ?
[21:43:33] * gdamore curses once again the folks who decided to ship our platform GUI tools as Tcl/Tk scripts.
[21:43:42] *** schily__ is now known as schily
[21:43:49] <jerome__> dclarke b49
[21:43:50] <jamesd_> gdamore, better than  java...
[21:44:01] <schily> Hi Dennis
[21:44:05] <gdamore> not when you have to build Tcl/Tk and BWidget everytime you do a new bug fix.
[21:44:05] <dclarke> jerome__: cool .. I have that here also
[21:44:11] <jerome__> bash-3.00# uname -a
[21:44:11] <jerome__> SunOS wfc-zit-ops-004 5.11 snv_49 i86pc i386 i86pc
[21:44:12] <dclarke> schily : hello !
[21:44:19] <dclarke> schily : you are on my hit list also
[21:44:32] <dclarke> jerome__: one sec while I create a new binary for you
[21:44:34] <jerome__> is there some dentry tunnable I could use?
[21:44:35] <jengelh> hit list as in slap?
[21:44:39] *** widjayy has joined #opensolaris
[21:44:42] <jamesd_> schilly is on many hit lists...
[21:44:44] <gdamore> so, i read somewhere that cdrecord in S10u2 (or u1?) can only do 2GB DVDs.  is that fixed?  in Nevada?
[21:44:45] <schily> you seem to produce strange problems now...
[21:44:56] <dclarke> jengelh: hit list as in "people I need to talk to"
[21:45:06] <dclarke> schily: thanks
[21:45:13] <dclarke> schily: but the facts are there
[21:45:14] <widjayy> hi
[21:45:19] <jerome__> dclarke what are you compiling?
[21:45:31] <dclarke> a small 16GB file load test
[21:45:35] <jerome__> hm ok
[21:45:41] *** widjayy has quit IRC
[21:45:56] <dclarke> it creates and then thrashes 62^3 files
[21:46:17] <jerome__> ah ok
[21:46:20] <jerome__> works for me
[21:46:23] <gdamore> its going to take a while, because dclarke created it as a 16GB .c file -- takes a while for the compiler to run... :-)
[21:46:36] <jerome__> heh
[21:46:42] <schily> dclarke: you send an impossible error message
[21:46:45] <dclarke> yeah right .. I unrolled all the loops
[21:47:12] <gdamore> (by hand)
[21:47:19] <jerome__> heh
[21:47:20] <jerome__> good one
[21:48:11] * gdamore sometimes wish i could get fixes backported from Nevada to S10u3.  like some of the keyboard/mouse fixes...
[21:48:13] <dclarke> # uname -a
[21:48:14] <dclarke> SunOS phobos 5.11 snv_49 i86pc i386 i86pc
[21:48:20] <dclarke> my build box .. one sec
[21:48:21] <jerome__> movr
[21:48:26] <jerome__> nice
[21:48:54] <dclarke> $ /opt/SUNWspro/bin/cc -V
[21:48:55] <dclarke> cc: Sun C 5.8 Patch 121016-02 2006/03/31
[21:48:55] <dclarke> usage: cc [ options] files.  Use 'cc -flags' for details
[21:48:59] <dclarke> Studio 11 okay ?
[21:49:10] <jerome__> I didn't install studio 11
[21:49:30] <jerome__> sorry
[21:49:33] <dclarke> install it
[21:49:34] <gdamore> dclarke is wonder if he can e'mail a 16GB binary.... :-)
[21:49:36] <dclarke> run it
[21:49:37] <dclarke> live it
[21:49:51] <dclarke> gdamore: all those loops compress nicely
[21:49:56] <gdamore> heh.
[21:50:12] <jerome__> I could
[21:50:19] <jerome__> but I'm tight in disk space
[21:50:32] <jerome__> how big is studio 11
[21:50:41] <dclarke> big
[21:50:44] <dclarke> 1GB
[21:50:47] <Error_404> gigantic
[21:50:49] <gdamore> 1-2 GB?
[21:50:51] <jerome__> damn
[21:51:00] <jerome__> I have 299Mb left
[21:51:04] <jerome__> on /
[21:51:06] <delewis> the fact that it includes a Java IDE doesn't help matters :-)
[21:51:09] <jerome__> I'd have to play tricks
[21:51:12] <delewis> or rather, includes Netbeans.
[21:51:31] <delewis> you could probably cut down the install footprint if you didn't bother installing it
[21:51:39] <dclarke> what will -xarch=?  be
[21:51:50] <gdamore> heh.  i'd love to see a minimal install that just included the C/C++ compilers, and dbx.  you can keep the rest of that crap.
[21:52:01] <delewis> gdamore: :-)
[21:52:08] <gdamore> s/install/distribution/
[21:52:09] <delewis> well, there's the math libraries, which are pretty cool :-)
[21:52:14] <delewis> and all of the tools are useful
[21:52:25] <gdamore> those are already in /usr/ccs. :-)
[21:52:26] <delewis> but I imagine that's < 250MB
[21:52:27] <jerome__> dclarke got binary?
[21:52:48] <gdamore> jerome__ got milk?
[21:52:49] <schily> dclarke: why do you have problems that cannot happen?
[21:52:52] <delewis> gdamore: nope
[21:53:01] <delewis> cscope, cxref, etc. aren't in /usr/ccs
[21:53:14] <gdamore> that's part of the extra crap.  i care about libm, and make.
[21:53:21] <delewis> ah, no wonder -- xemacs is also included
[21:53:29] <delewis> and gvim
[21:53:29] <gdamore> _with source_
[21:53:33] <delewis> yes
[21:53:52] <dclarke> schily : one minut please
[21:53:59] <dclarke> schily : I'll be right with you
[21:54:06] <delewis> $ du -sh xemacs-21.4.12
[21:54:07] <delewis>  285M xemacs-21.4.12
[21:54:10] <delewis> !
[21:54:13] <gdamore> schily: do you know if the 2GB problem in cdrecord is fixed in Solaris Nevada?
[21:54:28] <gdamore> (I don't have a Nevada system with a dvd burner right now)
[21:54:29] <schily> dclarke: ?????
[21:55:40] <schily> dclarke: there is no 2 GB problem and there has been no such "problem" since Y2000
[21:56:51] <gdamore> i wonder why it was listed as an issue for S10u1 then.
[21:57:19] <delewis> the only time I recall a DVD issue/2GB was in the Solaris 8 days (which schilly noted would be around Y2000)
[21:57:32] <delewis> we've got that mentioned in the MPlayer docs, particularly.
[21:57:33] <jerome__> is there any dentry setting I can tune for zfs?
[21:58:12] <gdamore> i figured it was a problem in the Solaris integration of cdrecord, since my local install of it works flawlessly.  but again, I don't have a nevada system that I can test it with right now.
[21:58:42] * delewis doesn't have Nevada installed on his only workstation with a DVD burner
[21:58:46] <delewis> otherwise, I'd test
[21:59:31] <gdamore> my only nevada system at the moment is a SPARCLE 500 laptop, which has only CDRW/DVD-ROM.  i'm trying to get a DVD-RW but they're hard to come by.
[21:59:42] <gdamore> (internal dvdrw, that is)
[21:59:54] <schily> well, there _was_ a problem on Solaris 8 because Sun did publish a very old mkisofs instead of a recent one, but this was in January 2000 and it was a problem caused by Sun
[22:00:03] <dclarke> bash-3.00$ file crucible
[22:00:05] <dclarke> crucible: ELF 64-bit LSB executable AMD64 Version 1 [SSE2 SSE FXSR FPU], dynamically linked, stripped
[22:00:17] <delewis> interesting
[22:00:24] <gdamore> i'm trying to find out where i saw the bug report.  hang on.
[22:00:30] <delewis> when did 'file' output extensions for x86?
[22:00:43] <dclarke> cool
[22:00:46] <gdamore> i think it has for quite a while.
[22:00:56] <delewis> I only recall it doing that for SPARC in the case of SPARCv8 a/b and SPARCv9 a/b (VIS 1 and 2)
[22:02:36] *** _william_ has joined #opensolaris
[22:02:40] <delewis> $ file testing
[22:02:41] <delewis> testing:        ELF 32-bit LSB executable 80386 Version 1, dynamically linked, not stripped, no debugging information available
[22:02:55] <gdamore> how did you compile it though?
[22:03:05] <delewis> gcc -march=i686 -msse
[22:03:09] <_william_> hi all
[22:03:20] <gdamore> okay, i stand corrected then. :-)
[22:03:24] <delewis> (client system, no Sun Studio installed) :-(
[22:03:31] <delewis> though, it shouldn't make a difference
[22:03:51] <delewis> what's the gcc -march tunable for EM64T?
[22:03:56] <delewis> prescott something?
[22:04:26] <jerome__> dclarke any luck?
[22:05:12] <dclarke> http://www.blastwave.org/dclarke/crucible/index.html
[22:05:19] <dclarke> goto that page
[22:05:19] <jengelh> delewis : no
[22:05:23] <jengelh> em64 is -mcpu=nocona
[22:05:28] <dclarke> jerome__: see bottom
[22:05:35] <dclarke> can you download that ?
[22:05:40] <delewis> $ gcc -march=prescott -msse -mfpmath=sse -msse2 -msse3 testing.c -o testing
[22:05:40] <delewis> $ file testing
[22:05:41] <delewis> testing:        ELF 32-bit LSB executable 80386 Version 1, dynamically linked, not stripped, no debugging information available
[22:05:58] <jerome__> dclarke ok
[22:06:03] <jengelh> delewis : so?
[22:06:44] <delewis> jengelh: so, 'file' isn't returning any extensions (though, it's possible this code is too simple and the compiler isn't using extensions)
[22:06:47] <jerome__> it's cooking
[22:06:48] <dclarke> jerome__: do you have 16GB of file system area to test with ?
[22:06:53] <jerome__> yes
[22:06:55] <dclarke> cooking ?
[22:06:55] <jerome__> easily
[22:07:00] <dclarke> okay .. cool
[22:07:02] <dclarke> ZFS ?
[22:07:12] <gdamore> schily: http://docs.sun.com/app/docs/doc/817-0547/6mgbdbsmb?a=view
[22:07:14] <dclarke> if you are on ZFS then you may get a warning
[22:07:25] *** gtc__ has joined #opensolaris
[22:07:27] <jerome__> I saw
[22:07:28] <dclarke> but it will run fine
[22:07:35] <jerome__> I didn't even know about the compression
[22:07:37] <jerome__> heh
[22:07:38] <gdamore> look for cdrecord.  this is where I found the complaint about 2GB and cdrecord.  S10u2 Whats New.
[22:07:38] <dclarke> do you have compression enabled ?
[22:07:45] <jerome__> how long does it take?
[22:07:53] <jerome__> dclarke no
[22:07:58] <dclarke> how fast is your system ? :-)
[22:08:01] <jerome__> dclarke I didn't even know zfs supports compression
[22:08:05] <jerome__> dual opteron 265 or 275
[22:08:14] <jerome__> 4 scsi dries on zfs in raid0
[22:08:15] <dclarke> ZFS with compression is generally faster than without
[22:08:21] <delewis> jerome__: of course it does -- in fact, compression usually offers better performance
[22:08:39] <delewis> provided you aren't CPU bound
[22:08:41] <dclarke> well .. its cranking out LOTs of files across a wide and not too deep directory structure
[22:08:52] <dclarke> 16GB woth of them
[22:08:55] * timeless frowns
[22:09:04] <dclarke> so we need to wait to get the reports
[22:09:08] <timeless> 5c3315ac ::print ire_t
[22:09:21] <delewis> jerome__: you might consider dropping zfs-discuss at opensolaris dot org an email about this, and let the ZFS gurus sort it out.
[22:09:24] <timeless> kmdb: failed to read ire_next pointer at 5c3315ac: no mapping for address
[22:09:29] <jengelh> delewis : no ,, what extension?
[22:09:35] <jengelh> .exe?
[22:09:51] <delewis> jengelh: nevermind.
[22:09:54] <dclarke> schily : did you see my build log that I posted ?
[22:09:57] <jerome__>  RT= 170.542584 sec
[22:10:02] <jerome__> 238328 files  avg=0.000715 sec  total=170.459894 sec  io_avg=87.384192 MB/sec
[22:10:07] <jengelh> No, `file` does not look at the ASM code, and hence, prints no -march extension
[22:10:28] <jengelh> In fact, you would need to thoroughly analyze the ASM to say whether it's -mpentium4 or -mathlon
[22:10:39] <delewis> jengelh: then why did dclark's 'file' output on his binary indicate it was compiled with sse, sse2, and sse3?
[22:10:45] <jengelh> or in short, `file` only looks at the ELF header
[22:10:58] <dclarke> jerome__: thats fast
[22:11:02] <jengelh> and the ELF header knows "x86" ("80386") or "x86_64"
[22:11:19] <dclarke> jerome__: I need to change the avg time for file write to scientific notation
[22:11:37] <jengelh> Actually, you can compile a program that makes use of both -march=pentium4 and -march=athlon, both -msse and -msse-msse2. File can't figure that out, and honestly, you can't either with objdump
[22:11:41] *** comay has joined #opensolaris
[22:12:01] <jamesd_> hi cory
[22:12:24] <delewis> jengelh: again, how did 'file' know dclarke's application had been compiled with sse, sse2, and sse3?
[22:12:46] <jerome__> dclarke if that's fast then what's my bottleneck?
[22:13:03] <jerome__> why I can't read() more than 150 files a second?
[22:13:13] <jerome__> in randomio
[22:13:15] <jengelh> delewis: gcc must have left a crater in the executable (mkisofs does that too, evil evil!)
[22:13:20] <schily> dclarke: yes, and it verifies that you compiled astoull.c
[22:13:29] <delewis> jengelh: interesting
[22:13:37] <schily> dclarke: so you should have atoullb() in libschily
[22:13:45] <gdamore> anyone know if its possible to get the s10u2 patch that delivers cdrecord seperately.  then i can test it on s10 3/05, which is where i have my dvdrw
[22:13:49] <dclarke> libschily ?
[22:13:55] <jerome__> dclarke: let's wait until this test is complete and I'll email you the results (I have to go to a meeting) could you give me your email @?
[22:13:55] <jengelh> delewis: Question is, what does "file" say when you compile foo.c without -msse, bar.c with -msse2, and then combine foo.o and bar.o to the final program.
[22:14:10] <dclarke> jerome__: dclarke at blastwave dot org
[22:14:14] <jerome__> dclarke: do you think having 40,000,000 files is an issue for zfs?
[22:14:33] <dclarke> jerome__: add six more zeros and then maybe
[22:14:34] <jerome__> do you think the directory entry cache is the bottleneck?
[22:14:37] <jerome__> heh
[22:14:38] *** bougie has quit IRC
[22:14:40] <schily> dclarke: use nm to check the file astoull.o
[22:14:41] <gdamore> jengelh: but SSE extensions should be in the ELF header. :-)
[22:14:48] <dclarke> schily : one sec
[22:14:49] <jengelh> gdamore: I don't disagree.
[22:15:13] <jengelh> gdamore: But what would `file mplayer` return then? "80386 maybe-sse maybe-sse2 maybe-sse3 maybe-3dnow?" ah thanks...
[22:15:46] <gdamore> well, i suppose it depends on how mplayer was compiled.  you don't _have_ to put the info in the ELF header, but you _can_
[22:15:49] <jengelh> well look at /etc/magic and check why it can print SSE
[22:16:11] <sommerfeld> need to distinguish "can-benefit-from-sse47" vs "needs-sse47-or-it-falls-over"
[22:16:20] <jengelh> my /etc/magic does not have any sse detection at all
[22:16:26] <dclarke> schily: what am I looking for here ?
[22:16:38] <dclarke> bash-3.00$ nm ./lib/OBJ/sparc-sunos5-cc/astoull.o./lib/OBJ/sparc-sunos5-cc/astoull.o:
[22:16:39] <schily> dclarke: so you should have atoullb() in libschily
[22:16:44] <gdamore> it doesn't use /etc/magic.  it uses elf(3ELF).
[22:16:56] <dclarke> there is no libschily here
[22:16:59] <dclarke> that I see
[22:17:02] <dclarke> let me look
[22:17:06] <jengelh> gdamore : That's interesting.
[22:17:17] <delewis> $ file `which mplayer`
[22:17:18] <delewis> /opt/local/bin/mplayer: ELF 32-bit MSB executable SPARC32PLUS Version 1, V8+ Required, UltraSPARC1 Extensions Required, dynamically linked, stripped
[22:17:18] <schily> dclarke: so you should have atoullb() in astoull.o
[22:17:19] <jengelh> maybe something IS in the header?
[22:17:21] <dclarke> ./libs/sparc-sunos5-cc/libschily.a
[22:17:23] <dclarke> that ?
[22:17:24] *** LeftWing has quit IRC
[22:17:25] <delewis> but that was compiled with v8plusb
[22:17:48] <delewis> er
[22:17:51] <delewis> v8plusa, rather
[22:18:06] *** mage2 has joined #opensolaris
[22:18:10] <mage2> Anyone home?
[22:18:14] <jengelh> delewis: Try `gcc -c -o main.o main.c; gcc -o program -msse -msse2 -msse3 main.o` and tell the result of `file`
[22:18:14] <mage2> :)
[22:18:43] <jengelh> as well as the reverse, `gcc -c -o main.o main.c -msse -msse2 -msse3; gcc -o program main.o`
[22:18:58] <jerome__> test 2
[22:19:03] <jerome__> RT= 560.107645 sec
[22:19:05] *** Netwolf has quit IRC
[22:19:08] <jerome__> 238328 files  avg=0.002349 sec  total=559.905743 sec  io_avg=0.831362 MB/sec
[22:19:14] <gdamore> i don't know about with gcc, but it is _possible_ to have assembler even for V8a, v8aplus, etc. that isn't declared in the ELF header.
[22:19:38] <gdamore> the ELF header just helps the linker decide whether the object is compatible or not.
[22:19:49] <dclarke> schily : see bottom of http://www.blastwave.org/dclarke/stuff/star_build.log
[22:19:54] *** LeftWing has joined #OpenSolaris
[22:20:09] * delewis just looks at file.c source
[22:20:10] <delewis> :-)
[22:20:36] <mage2> I have what I think is a newbie question. I have solaris on a system and have upgraded the nic driver software with a pkg from broadcom to fix a bug in the inital release. Now if I want to uninstall those drivers (bcme) and go back to the bge drivers. The way I do this is down the interfaces, unplub them, remove the pkg.
[22:20:47] <mage2> Now how do i enable the old drivers ?
[22:21:08] <sommerfeld> pkgadd the SUNWbge package
[22:21:24] <sommerfeld> (did this recently on a ferrari 4k)
[22:21:40] <delewis> still no wireless support on the Ferrari 4k :-(
[22:21:48] <mage2> Thats what I thought. Anyidea where that blasted pkg is? I looked around and felt I was missing something
[22:22:13] * mage2 is not a solaris admin and knows it
[22:22:26] <sommerfeld> how did you install this beast originally?
[22:22:31] <sommerfeld> boot from DVD?
[22:22:36] <mage2> cd or dvd
[22:22:37] <gdamore> schily: did you look at the URL i posted for the problem with cdrecord/2GB?
[22:22:38] <delewis>  if (flags & EF_SPARC_HAL_R1)
[22:22:38] <delewis>                                 (void) printf("%s", gettext(
[22:22:38] <delewis>                                     ", HaL R1 Extensions Required"));
[22:22:38] <mage2> cd i think
[22:22:39] <delewis> :-)
[22:23:35] <mage2> sommerfeld CD
[22:23:40] <sommerfeld> there will be a directory on the CD with lots of subdirs, one per package
[22:23:46] <sommerfeld> each CD, really
[22:24:18] <sommerfeld> SUNWbge should be on one of the cd's
[22:24:28] <mage2> Hmm so the best option is just find the pkg on the cd? Or is there a online place i could get it from?
[22:24:56] <delewis> mage2: pkgs on the media are not re-distributable.
[22:24:57] <mage2> ok
[22:24:58] * timeless grins
[22:25:04] <timeless> debugging solaris kernel is not bad
[22:25:04] <sommerfeld> the only way to download at present is complete cd images
[22:25:10] *** ProfMikey has quit IRC
[22:25:12] <mage2> i was afraid of that
[22:25:15] <mage2> ok thanks
[22:25:16] *** ProfMikey has joined #opensolaris
[22:25:23] <mage2> I will start searching CD's
[22:25:32] <mage2> Thank you all for your help .
[22:27:40] <sommerfeld> no problem
[22:28:30] <sommerfeld> delewis: so ferrari 4k has a nicely accessible mini-PCI slot and decent built-in antennae.  all oyu need is a atheros-based CM9 mini-pci card and you're all set.
[22:29:04] <sommerfeld> (and a small screwdriver..)
[22:29:15] <delewis> sommerfeld: yeah, I've done a bit a research, but it's still a bit of a hassle. Have you heard anything about hardware support on the Ferrari 5k?
[22:29:21] <delewis> (in general)
[22:29:55] <dclarke> schily : where did you run away too ?
[22:29:56] <sommerfeld> no, but i haven't been paying attention to laptop chatter recently since my 4k has been behaving nicely (except when the cat sits on the keyboard...)
[22:30:09] <delewis> sommerfeld: :-)
[22:30:18] <delewis> yeah, I'm looking forward to getting my Ferrari.
[22:32:04] <dclarke> okay .. question time
[22:32:22] <jengelh> okay .. answer time
[22:32:23] <dclarke> how does one provide RAID 1+0  via SVM ?
[22:32:32] <dclarke> I think that ZFS does this easily
[22:32:44] <dclarke> but what about Solaris Volume Manager ?
[22:33:02] <dclarke> RAID 1+0 == mirrors that are then striped
[22:33:03] <jamesd_> see the wonderful docs at  docs.sun.com
[22:33:09] <dclarke> baloney
[22:33:19] <dclarke> I have the SVM manual in front of me
[22:33:24] <dclarke> it glosses over the idea
[22:33:29] <dclarke> with no specifics
[22:33:50] <dclarke> but I can point specifically to page 83 of the manual where it refers to RAID 0+1
[22:33:51] *** k2flag has left #opensolaris
[22:34:08] <dclarke> and to RAID 1+0 but never actually gets around to ever really talking about the tough one
[22:34:27] <dclarke> its not like I'm new to the SVM game
[22:34:35] <jamesd_> create the mirrors, and then  create a concat of the devices.
[22:34:45] <dclarke> oh really ?
[22:35:07] <jamesd_> that is my guess..
[22:35:27] <dclarke> well its a good guess
[22:35:33] <dclarke> but the concat is not a stripe
[22:35:56] <jamesd_> then stripe instead of concat
[22:36:05] <dclarke> and metainit(1) is real clear at the bottom
[22:36:16] <dclarke> Stripes, concatenations, and RAID level 5  metadevices  must consist of slices only.
[22:37:41] <timeless> interesting\
[22:37:57] <timeless> so dtrace_probe2 is a nop slide when it isn't active? :)
[22:38:35] <jamesd_> dclarke, http://blogs.sun.com/andresblog/date/20050614
[22:39:44] <dclarke> jamesd : you have far better Google skills than me
[22:39:49] <dclarke> thats seems bang on
[22:40:03] <dclarke> and when I search for terms I get boatloads of crap
[22:40:07] <dclarke> let me read that
[22:40:10] <dclarke> thanks !
[22:40:15] *** krozinov has quit IRC
[22:43:35] <sommerfeld> dclarke: so, SVM wants you create a 0+1 (mirror of stripes); if the two stripes are built of disks of the same sizes it will cope with failures as if you'd build a 1+0 (stripe of mirrors)
[22:43:57] <dclarke> bloody hell
[22:44:17] <dclarke> not very well documented
[22:44:22] <dclarke> but okay .. soo
[22:44:40] <dclarke> if I have a boatload of Sun Firmware FCAL disks in a fibre array
[22:44:43] <dclarke> all identical
[22:44:57] <dclarke> perfectly the same X-option part number FRU number or whatever its called
[22:45:14] <dclarke> and then another boatload ( 22 of them ) FCAL disks all identical also
[22:45:17] <dclarke> on another array
[22:45:27] <dclarke> I can stripe the first set
[22:45:32] <dclarke> stripe the second set
[22:45:35] <dclarke> mirror
[22:45:48] <dclarke> and I get effectively RAID 1+0 ?
[22:45:57] <dclarke> is that the actual promise here ?
[22:46:20] <sommerfeld> yes
[22:46:26] <jamesd_> not sure... this is giving me a headache...  off for a walk
[22:46:40] <dclarke> jamesd : it gives me an ulcer
[22:46:52] <dclarke> sommerfeld: thanks for the insights
[22:47:00] <dclarke> I need to actually test this
[22:47:08] <dclarke> because I don't believe it really
[22:47:31] <dclarke> and I'll need to yank some disks out of the arrays to see if I can really sustain multiple failures
[22:48:03] <timeless> erm
[22:48:09] <sommerfeld> should survive multiple failures so long as both copies of the same data don't disappear.
[22:48:22] <timeless> ip`ire_ftable_lookup+0x231: movl $0x100,%eax
[22:48:26] <timeless> [1]> ::step
[22:48:27] <dclarke> sommerfeld: exactly the desired config !
[22:48:40] <timeless> int cannot be stepped - operation not supported by target
[22:48:42] <timeless> ??
[22:48:51] <dclarke> sommerfeld: I'll need to then sort out the UFS filesystem specifics also
[22:49:08] <dclarke> sommerfeld: the application that will run here is called "interwoven"
[22:49:25] <sommerfeld> two A5200's?
[22:49:26] <dclarke> sommerfeld: it costs nearly $32K per year per socket !
[22:49:35] <dclarke> sommerfeld: yep
[22:49:43] <dclarke> sommerfeld: I love A5200's
[22:49:51] <sommerfeld> i've got two of them here as a zfs pool
[22:49:55] <dclarke> sommerfeld: the "22 disks" is a dead give away
[22:50:05] <dclarke> sommerfeld: I tried to push ZFS
[22:50:10] <dclarke> I tried .. I really did
[22:50:33] <dclarke> but they were very very much the sceptics
[22:50:40] <sommerfeld> redundant zfs configs make A5200's much more reliable
[22:50:49] <dclarke> my point to them exactly
[22:51:02] <dclarke> but they would not even want to talk about Solaris 10 or ZFS
[22:51:14] <dclarke> I had to go to Solaris 9 Update 8
[22:51:20] <dclarke> silly buggers
[22:51:21] <delewis> ugh.
[22:51:34] <delewis> so much for Solaris' ABI stability if vendors aren't going to put faith in it.
[22:51:37] <dclarke> some IT directors should be spanked
[22:51:56] <delewis> they've got a Windows/Linux mindset -- you upgrade, you get biten.
[22:51:57] <delewis> bitten*
[22:52:09] <schily> dclarke: so everything is OK and astoullb is not missing
[22:52:26] <dclarke>  /dev/md/rdsk/d9:        923769306 sectors in 319754 cylinders
[22:52:43] <schily> gdamore: a wrong information from Sun
[22:52:48] <dclarke> schily: everything is NOT okay
[22:53:04] <dclarke> schily: star will not compile
[22:53:20] <dclarke> so .. you tell me whats wrong ?  the Makefile ?
[22:53:42] <schily> dclarke: ad you have all that is needed, everything is OK
[22:54:03] <dclarke> I repeat ... becuase you don't listen too well .. it is not all okay
[22:54:16] <dclarke> if all were "okay" then star would build
[22:54:20] <dclarke> it does not
[22:54:25] <dclarke> how is that "okay" ?
[22:54:33] <schily> dclarke: except when you have a self made LD_LIBRARY_PATH and link against a old libschily from somewhere else
[22:54:34] <dclarke> its bloody frustrating
[22:54:46] <dclarke> there is NO LD_LIBRARY_PATH !!
[22:54:53] <dclarke> dammit .. thats bloody obvious
[22:55:01] <dclarke> thats the first thing I get rid of
[22:55:12] <delewis> what about a path in the run-time linker configuration?
[22:55:35] <dclarke> that could be an issue
[22:55:39] <dclarke> however ...
[22:55:52] <dclarke> I am using the Studio 11 compiler from OpenSolaris.org
[22:56:00] <dclarke> the tarball that I extract
[22:56:07] <dclarke> my PATH is dead simple
[22:56:12] <delewis> that's not relevant
[22:56:15] <dclarke> LC_ALL=C and LANG=C
[22:56:17] <delewis> I'm referring to 'crle'
[22:56:19] <dclarke> nothing fancy
[22:56:25] <delewis> that run-time linker configuration
[22:56:40] <delewis> $ crle
[22:56:40] <delewis> Default configuration file (/var/ld/ld.config) not found
[22:56:40] <delewis>   Default Library Path (ELF):   /lib:/usr/lib  (system default)
[22:56:41] <delewis>   Trusted Directories (ELF):    /lib/secure:/usr/lib/secure  (system default)
[22:56:44] <delewis> that sort of thing.
[22:57:01] <dclarke> well .. before I get to dealing with crle I will deal with my A5200's
[22:57:06] <dclarke> then I'll be right back
[22:57:11] <delewis> check to see if you have a path in there that contains an old libschilly
[22:57:13] <dclarke> star is important to me
[22:57:18] <dclarke> bloody criticla in fact
[22:57:36] <schily> dclarke: I never had such problems
[22:57:45] <dclarke> I _know_ the problem
[22:57:51] <dclarke> I think I see the issue !
[22:57:56] <dclarke> I think
[22:58:01] <schily> dclarke: what is it?
[22:58:01] * dclarke guesses wildly
[22:58:21] <dclarke> I neet to remove the star 1.75 build from CSW
[22:58:32] <dclarke> the current CSWstar may be an issue
[22:58:38] <schily> dclarke: why?
[22:58:43] <dclarke> along with schilyutils
[22:58:48] <schily> dclarke: why?
[22:58:57] <dclarke> well .. like I said .. I'm guessing wildly here
[22:59:01] <schily> dclarke: I did not yet install shared libraries
[22:59:24] <schily> dclarke: or did I?
[22:59:46] <dclarke> $ crle
[22:59:46] <dclarke> Default configuration file (/var/ld/ld.config) not found
[22:59:46] <dclarke>   Platform:     32-bit MSB SPARC
[22:59:46] <dclarke>   Default Library Path (ELF):   /lib:/usr/lib  (system default)
[22:59:46] <dclarke>   Trusted Directories (ELF):    /lib/secure:/usr/lib/secure  (system default)
[22:59:57] <dclarke> let me try this
[23:00:01] <timeless> can anyone here suggest where in http://cvs.opensolaris.org/source/xref/on/usr/src/uts/common/inet/ip/ip_ire.c#4831 i'm likely to crash? :)
[23:00:08] <dclarke> let me pkgrm CSWstar and CSWschilybase
[23:00:10] <timeless> i guess i only have two choices
[23:00:19] <dclarke> and CSWschilyutils
[23:00:21] <timeless> recv_ill = 0x0, xmit_ill = 0x0
[23:00:25] <dclarke> and CSWcdrtools
[23:00:33] <dclarke> then try smake again
[23:00:37] <dclarke> one moment please
[23:00:45] <dclarke> I really _ned_ this star to work
[23:00:54] <dclarke> thats "need"
[23:01:06] * dclarke thinks tar sucks
[23:01:16] * dclarke cpio sucks tar
[23:01:22] <delewis> gtar > star :-)
[23:01:23] * delewis ducks
[23:01:35] <dclarke> gtar is crap !!  compared to star
[23:01:37] <jengelh> in terms of option parsing, yes
[23:01:45] <delewis> dclarke: you missed the [sic], apologies.
[23:01:53] <jengelh> looking for -cvjf in star? Wrong dude....
[23:01:54] *** eboutilier has joined #opensolaris
[23:01:58] <schily> dclarke: this cannot create problems as I even have /opt/schily which is in the makefiles and I get no problems
[23:02:02] <delewis> I prefer cpio over tar implementations -- greater granularity in file selection in the archive.
[23:02:11] <schily> ???
[23:02:13] <jengelh> delewis : You just don't knwo about --no-r in tar!
[23:02:14] <gdamore> schily: thanks.  i'm looking forward to getting a DVD RW drive so I can verify it.
[23:02:17] <oxygene> delewis: star contains sfind
[23:02:21] <jengelh> (--no-r = maximum file granularity)
[23:02:28] <schily> cpio is worse that star
[23:02:43] <jengelh> star should accept gtar options
[23:02:47] <schily> do you have find(1) in cpio?
[23:02:57] <delewis> schily: how is star integration in Solaris proceeding?
[23:02:58] *** sahafeez has quit IRC
[23:03:01] <jengelh> find what?
[23:03:06] <jengelh> why would one need find in cpio?
[23:03:12] <delewis> schily: you don't need find in cpio -- that's the point.
[23:03:19] <schily> star was planned to appear in /usr/bin on Solaris 10
[23:03:19] <dclarke> look .. star is so damn perfect at creating portable archives from system to system that it beats all the others .. hands down
[23:03:24] <jengelh> find in cpio, that's the most blatant rule to violate the unix philosophy
[23:03:25] <delewis> cpio doesn't find files -- it leaves that aspect of choice to you.
[23:03:44] *** laca is now known as lacaAFK
[23:03:46] <schily> this is a common missconception of the UNIX rules
[23:03:57] <jengelh> find / -iname '*.iso' -print0 | tar -T- --null --no-r -cvf /tmp/foo.tar
[23:04:00] <jengelh> DID I MISS SOMETHING
[23:04:13] *** Mazon is now known as mazon
[23:04:16] <delewis> dclarke: I should hope that the Solaris tar creates portable archives
[23:04:16] <schily> star -find allows you to do things you can't do with sparate utilities
[23:04:18] <dclarke> CFLAGS=-xstrconst -xildoff -xarch=sparcv7 -H -Kpic -xlibmieee -xlibmil -xtime -Xa
[23:04:28] <dclarke> okay schily ?
[23:04:31] <jengelh> schily : example plz
[23:05:07] <schily> star -x -find -type d -chown root -chmod g+w
[23:05:24] <dclarke> delewis: Sun tar is anything but perfect
[23:05:30] <jengelh> apart from the -chmod thing, I can do that with external tools
[23:05:32] <delewis> and why does that over an advantage over using 'find' and piping the output to your archiver of choice?
[23:05:44] <jengelh> find . -type d -print0 | tar -T- --null --no-r --owner=root -c
[23:05:47] <delewis> (apart from the chmod, that is)
[23:05:51] <jamesd_> dclarke, did you do tuning for io on your box with the disk array... by default solaris only reads in  128K chunks well it did pre  solaris 10
[23:05:56] <jamesd_> not sure about now
[23:06:00] <schily> you can't do this with EXRACTION
[23:06:01] <jengelh> can't see why find+tar could not create the archive the same way
[23:06:10] <jengelh> for extraction use mc
[23:06:11] <jengelh> :p
[23:06:14] * jengelh evades
[23:06:22] <timeless> single-step stop on page fault (#pf) trap
[23:06:23] <timeless> ip`ire_local_same_ill_group+0x1c: movq 0x28(%r9),%r9
[23:06:23] <timeless> %r9 = 0
[23:06:23] <schily> you cannot use find in extract/find/diff mode
[23:06:30] <delewis> dclarke: with respect to ACLs, perhaps, but it does create portable archives.
[23:06:31] <dclarke> jamesd : yeah .. I made adjustements in the /etc/system .. good call too
[23:06:38] <jengelh> schily : If anybody _needed_ that functionality it would have been long in the UNIX tars
[23:06:41] * timeless sighs
[23:07:06] <schily> you cannot chown/chmod in create mode and you need twice as many stat calls
[23:07:07] <oxygene> jengelh: if anybody needed -z or -j in tars, it would be in the posix spec, right? ;)
[23:07:17] <delewis> "tar wars"
[23:07:19] * delewis hides
[23:07:27] <schily> If there was libfind before I created it, it would be in UNIX tar long ago
[23:07:31] <jengelh> oxygene: standardization efforts are slowe than programming efforts, sadly
[23:07:36] <gdamore>  /brick delewis
[23:08:06] <gdamore> and ARC processes are slower still. :-)
[23:08:27] * timeless is confused
[23:08:38] <dclarke> I'll say this for star .. I have been in situations when I needed to move 500GB and more of millions of files from system to system and nothing did the job with total confidence other than star
[23:08:52] *** widjayy has joined #opensolaris
[23:08:58] * jengelh rsyncs
[23:08:59] <schily> jengelh: you need to learn what to do with star -find and you will like it
[23:09:08] <dclarke> in one case I moved email for 13,000 users
[23:09:13] <delewis> "total confidence" sounds like an odd concatenation of words spun off by marketing departments.
[23:09:16] <delewis> in the case of tar that is
[23:09:26] <delewis> all that should matter is the tar implementation in question creates portable archives
[23:09:29] <delewis> the Solaris tar does
[23:09:30] <dclarke> delewis: did you just call me a marketing person ?!?!
[23:09:36] <delewis> that's "total confidence" for you.
[23:09:38] <dclarke> delewis: did ya ?
[23:09:51] <delewis> dclarke: :-)
[23:09:55] * dclarke thems fighting words
[23:10:15] <dclarke> :-P
[23:10:20] <delewis> it's a bit of an overstatement to say that you had "total confidence" in 'tar', though. :-)
[23:10:37] <jengelh> schily : Wonderful. Make a SUSE package.
[23:10:55] <schily> BTW: Did someone compile and test the latest solaris with Joliet support?
[23:11:03] <dclarke> I will now start with a fresh build of smake-1.2a38
[23:11:10] <delewis> schily: it isn't in 20061009 is it?
[23:11:11] *** lloy0076 has joined #opensolaris
[23:11:47] <delewis> I don't think it was as the #defition NAME_LEN (or whatever) still appears to be 64 chars
[23:11:54] <timeless> WTF
[23:12:13] <timeless> does anyone here do any solaris kernel hacking?
[23:12:25] *** Tpenta has joined #opensolaris
[23:12:36] <schily> since today, I have a hsfs that does everything correct: root dir metadata and hardlinked files finally work as expected, sou you may backup your root filesystem with mkisofs
[23:12:40] <dclarke> morning Tpenta
[23:12:54] <delewis> schily: does hsfs finally support > 64 character filenames?
[23:13:00] <delewis> without rock-ridge that is
[23:13:04] <dclarke> there is a good question
[23:13:18] <dclarke> the Blastwave DVD has files longer than 64 chars
[23:13:30] <schily> hsfs supports now 110 Joliet unicode characters. This may be 330 UTF-8 bytes with katakana and will cause Solaris higher FS layers to fail
[23:13:34] <dclarke> on Windows it looks wrong
[23:13:38] <delewis> dclarke: which is fine, but you're probably using Rock-Ridge/Joliet
[23:13:55] <delewis> ISO1999 supports > 64 characters, meaning you don't have to use an extension
[23:14:06] <delewis> schily: excellent!
[23:14:27] <delewis> you said Joliet, though :-(
[23:14:27] <timeless>    2510 } else if (ire->ire_rfq != NULL) {
[23:14:28] <timeless>    2511 ill = ire->ire_rfq->q_ptr;
[23:14:36] <schily> The latest OSOL also includes ISO-9660:1999 support
[23:14:40] * timeless sighs
[23:14:43] * timeless needs to wake up
[23:14:45] <delewis> ah, that's what I want
[23:15:07] <delewis> unfortunately, 20061016 isn't on genunix, yet
[23:15:21] *** widjayy has quit IRC
[23:16:45] <schily> for the new "correct hard link" support you will need an updated mkisogs
[23:16:48] <schily> for the new "correct hard link" support you will need an updated mkisofs
[23:17:42] *** sahafeez has joined #opensolaris
[23:18:01] <schily> unfortunately, testing/code-review and burocracy for integration rights take some time for Solaris, so I don't expect the "correct inode/hard-link" support in less than a months
[23:18:52] *** ChanServ sets mode: +o Tpenta
[23:19:07] <jengelh> hah it's also lacking in solaris?
[23:19:28] <schily> what's lacking?
[23:19:35] <jengelh> correct inode'ing
[23:20:00] <schily> this is wrong on all OS today and Solaris is the first OS to fix this.
[23:20:21] <jengelh> depends
[23:20:36] *** kimc has joined #opensolaris
[23:20:39] <schily> ??? Linux does it wrong, BSD does it wrong
[23:21:19] <jengelh> and solaris does it wrong you said
[23:21:54] <schily> yes, the other OS do it wrong because they did just copy the inode algorithm from Solaris from 1990 ;-)
[23:22:56] <jengelh> so solaris is the root of all evil^H^H^H^Hthat
[23:23:11] <Seawolf99> need help upgrade from bfu. I get this message "Run /ws/onnv-gate/public/bin/migrate_bind9 to migrate to BIND 9" while upgrading from build_46 to the current.
[23:23:17] <schily> Solaris is the root of innovation....
[23:23:33] <jengelh> o rly?
[23:23:47] <jengelh> why do i still miss a reiserfs, jfs and xfs driver for solaris?
[23:24:11] <AbeFroman> aix, irix, and one guy
[23:24:17] <jengelh> heh
[23:24:22] <delewis> given how hard of a time ReiserFS had making it into the Linux kernel, it would never make Solaris QA standards in our lifetimes.
[23:24:22] <schily> because you have ZFS and do not need that "murder" FS
[23:25:04] <sickness> hi schily :)
[23:26:04] <schily> note that reiserfs quality even causes headaches for Linux people
[23:26:09] <dclarke> $ smake --version
[23:26:09] 
[23:26:14] <dclarke> looks perfect thus far
[23:26:19] <delewis> schily: of course
[23:26:32] <Tpenta> good afternoon mr clarke
[23:26:33] <jengelh> schily : Yes, now. It's dated. Once was state of the art.
[23:26:34] <schily> your problem is not caused by smake
[23:26:35] <delewis> XFS had a lot of trouble too getting included
[23:26:40] <delewis> I don't recall much about JFS
[23:26:51] <jengelh> and gfs2 and ocfs, not to mention unionfs and squashfs
[23:27:02] <dclarke> can't hurt to rebuild it AFTER I removed all the CSWstar related packages
[23:27:18] <jamesd_> delewis, yes.. but  at least it was safe... it was a prodution ready filesystem.... reisfs is not.
[23:27:45] <schily> I am away for some minutes
[23:27:51] *** sommerfeld has quit IRC
[23:27:58] <delewis> jamesd_: definitely -- XFS only violated Linux kernel development practices, while ReiserFS (and Reiser4) not only violate that but quality control standards (by Linux standards)
[23:28:26] <Tpenta> Linux has kernel development practices? [ heavily implied :-) ]
[23:28:36] <delewis> Hans Reiser has said some things that made me cringe regarding how much importance he places on filesystem stability.
[23:28:39] <sickness> if I remember from an interview to andrea arcangeli, xfs had a lot of "overlapping" things with the linux vfs layer wich in irix did xfs but in linux were already implemented in the vfs so it had to be sorta "checked" more than other filesystems for proper integration...
[23:28:39] <jengelh> Of course, Documentation/CodingStyle on one side
[23:28:45] * dclarke watches star build .. again ..
[23:28:51] <delewis> sickness: yes
[23:28:53] *** Netwolf has joined #opensolaris
[23:28:54] <jamesd_> delewis, linux has   quality control standards? i thought the only test was that it built semi-cleanly and not blow up on boot.
[23:28:56] <dclarke> be back in a bit .. need air
[23:29:00] *** jmcp has quit IRC
[23:29:05] <delewis> the filesystem Linux kernel developers want you to use the Linux kernel vfs layer
[23:29:26] <delewis> initially, the SGI team implemented their own vfs layer, but later integrated their changes into the Linux one, IIRC.
[23:29:37] <delewis> Hans didn't want to comply, period, stating his layer was better in some ways.
[23:29:58] <jengelh> Interestingly, we also hear Schily that solaris device addressing is also better than linux's
[23:30:04] <delewis> jamesd_: I was being *heavily* sarcastic, hence the "even by Linux kernel standards"
[23:30:05] <jengelh> same problem different ppl
[23:30:33] <delewis> though, there's at least some quality control if they're unwilling to integrate Reiser4 in it's current form.
[23:30:36] <delewis> :-)
[23:31:38] <sickness> well, I see also a lot of "politics" in linux...
[23:32:16] <sickness> maybe now that novell/suse are about to drop it, reiserfs will lack a big "sponsor" and could be kicked out of the vanilla tree...
[23:32:22] <delewis> most of the arguments I've seen for not integrating Reiser4 in it's current state seem to be valid.
[23:32:29] <jengelh> sickness : unlikely
[23:32:32] <delewis> Hans has duplicated a lot of the Linux VFS layer, making it difficult to maintain
[23:32:48] <jengelh> sickness : ADFS has not gone away either
[23:32:51] <delewis> not to mention he's violating standards outlined by the Linux coding standards "document"
[23:32:58] <jengelh> delewis : everyone violates that mostly
[23:33:03] <jengelh> more or less
[23:33:12] <delewis> jengelh: interesting.
[23:33:18] <delewis> they seem to be really pressing it in the case of Reiser4
[23:33:21] <jengelh> fs/xfs does not always adhere to it
[23:33:23] <mage2> ok so apparently the package SUNWbge is a ghost
[23:33:31] <jengelh> even gfs2 and unionfs needed a swift kick
[23:33:56] <jengelh> i'm not even talking about Stupid C that is crumbling everywhere [useless casts, for one]
[23:33:56] <delewis> jengelh: yes, but I think the xfs developers are willing to comply most of the time
[23:34:00] <delewis> Hans isn't :-)
[23:34:16] <jengelh> delewis: Yes they are. Over time, the IRIX fluff will be removed
[23:34:17] *** rltns has joined #opensolaris
[23:34:27] <delewis> but what I know of the current filesystem situation in Linux comes from 3rd parties and glancing over LKML when I have nothing better to do.
[23:34:33] <sickness> jengelh: well, adfs is another thing :)
[23:34:42] <jengelh> sickness : It's the same thing as you say.
[23:34:47] <sickness> jengelh: and maybe reiser3.6 will not go away too...
[23:34:49] <jengelh> No one uses it, yet it's not removed.
[23:34:59] <jengelh> and there are definitely more reiser3 users than adfs
[23:35:08] *** gtc__ has quit IRC
[23:35:18] <sickness> yeah, that makes sense
[23:35:38] <jengelh> arch/ probably has one or two directories solely for less than 100 ppl
[23:35:51] *** rltns has quit IRC
[23:40:05] *** comay has quit IRC
[23:41:05] *** lloy0076 has left #opensolaris
[23:42:28] <mage2> anyone know where SUNWbge hides on the cd install media
[23:42:29] <mage2> ?
[23:44:06] <mage2> the package that is
[23:44:58] <elektronkind> Solaris_1?/Product
[23:45:09] <jamesd_> i would guess disk1.. since it would obviously be part of the core install
[23:45:12] <elektronkind> where ? is either 0 or 1, depending
[23:45:23] <mage2> i looked and im missing it somehow
[23:45:28] <mage2> i dont just ask dumb questions
[23:45:29] <mage2> hehe
[23:45:36] <mage2> i go look again
[23:45:42] <elektronkind> have you checked all the CDs?
[23:45:45] <mage2> yes
[23:45:47] <mage2> thats why im here
[23:45:58] <elektronkind> wait, SUNWbge
[23:46:17] <Seawolf99> need help. I am trying to upgrade from snv_46 to the most recent bfu archive. but I get a message saying BIND 8 is not in ON anymore, migrate to BIND 9 and it aborts
[23:46:30] <Seawolf99> where can I find the migrate script
[23:46:59] <elektronkind> mage2: yeah, should be on one of them.
[23:47:09] <mage2> i am trying to reinstall the blasted default drivers
[23:47:16] <mage2> hehe
[23:47:40] <elektronkind> the "which CD is this package on" game is why I got a DVD burner :)
[23:48:01] <mage2> haha
[23:48:14] <mage2> i agree
[23:48:24] <mage2> blasted 5 cd installs
[23:50:03] <mage2> ok
[23:50:14] <mage2> im going to search somemore
[23:56:11] *** gisburn has joined #opensolaris
[23:56:33] * gisburn looks for sickness
[23:56:41] <sickness> hi gisburn :)
[23:56:41] <gisburn> !seen sickness
[23:56:42] <Drone> sickness is currently online in #opensolaris and last spoke on Wed 18 Oct 2006 21:57 GMT, saying 'hi gisburn :)'.
[23:56:48] <gisburn> eh
[23:56:51] <gisburn> heh
[23:56:53] <gisburn> sickness: Hi!
[23:57:10] <sickness> I'm really in a hurry of bandwidth, I don't think that you will be happy to work via ssh this evening :/
[23:58:01] <gisburn> sickness: is the new SG1 episode out ? Or SG/A ?
[23:58:28] <sickness> what's sg1? :)
[23:58:46] <gisburn> sickness: StarGate SG-1 and StarGate Atlantis
[23:59:10] <sickness> oh, I don't know :/





top