[00:05:09] <ProfMikey> oh this is #opensolaris
[00:05:11] <ProfMikey> silly me
[00:05:17] <icon> *blink*
[00:05:50] <Odin-> ProfMikey: ... but not what?
[00:06:07] <ProfMikey> looks like I ran out of the alco in my blood streem
[00:06:35] <ProfMikey> Odin-: pgup, thought this is linux chan
[00:07:08] <Odin-> ProfMikey: Heh. :p
[00:07:17] <ProfMikey> yeah, heh indeed
[00:07:18] <Odin-> "Whoops."
[00:07:28] <ProfMikey> Whoops, I went too far
[00:09:48] <rawtaz_> aha :)
[00:10:05] <ProfMikey> channel here, channel there, channel everyhere.... more confusion heh... like IBM... "Innovation For Profit" I would have said "Complexity For Confusion"
[00:13:39] <icon> Complexity for Personal Gain more like it
[00:14:05] <ProfMikey> that's your version :)
[00:31:36] *** kloczek has quit IRC
[00:31:46] *** _william_ has quit IRC
[00:33:24] *** svoboda has quit IRC
[00:34:50] *** Mazon is now known as mazon
[00:39:11] *** _william_ has joined #opensolaris
[00:39:40] *** coffman has joined #opensolaris
[00:45:19] *** Fish has quit IRC
[00:46:45] *** svoboda has joined #opensolaris
[01:01:53] *** kloczek has joined #opensolaris
[01:05:41] *** solarisjon has quit IRC
[01:09:51] *** B|nTaRa has quit IRC
[01:18:19] *** B|nTaRa has joined #opensolaris
[01:18:23] <B|nTaRa> .j gnusol
[01:18:35] <B|nTaRa> opss
[01:18:40] *** dwc- has quit IRC
[01:21:53] *** stevel has quit IRC
[01:25:23] *** stevel has joined #opensolaris
[01:25:23] *** ChanServ sets mode: +o stevel
[01:30:31] *** polk__ has quit IRC
[01:31:39] *** polk__ has joined #opensolaris
[01:37:31] *** sommerfeld has quit IRC
[01:42:27] *** deather_ has joined #opensolaris
[01:56:07] *** piwi has joined #opensolaris
[02:00:15] *** deather has quit IRC
[02:03:52] *** _william_ has quit IRC
[02:10:33] *** solaris-user has quit IRC
[02:11:48] *** Sieghard has quit IRC
[02:12:59] *** Sieghard has joined #opensolaris
[02:20:41] *** vmhobbes has quit IRC
[02:20:47] *** Sieghard has quit IRC
[02:31:29] <stevel> anyone here have old copies of ON nightlies? like maybe 20060201 or 20060208?
[02:36:04] *** coffman has quit IRC
[02:37:38] *** koolniczka has quit IRC
[02:37:52] *** alanc-away is now known as alanc
[02:38:03] *** b_e_n_z has joined #opensolaris
[02:40:57] *** Tpenta has quit IRC
[02:41:58] *** Tpenta has joined #opensolaris
[02:42:46] *** ShadowHntr has joined #opensolaris
[02:43:38] *** lacaAFK is now known as laca
[02:46:32] *** gisburn has joined #opensolaris
[02:47:21] <gisburn> <rant mode="turbochicken"> The /topic is so... use-l-e-s-s ... </tant>
[02:47:34] <gisburn> s/tant/rant/
[02:48:20] * gisburn adds something to his xmas wishlist: two new hands which do not write so many typos.
[02:48:47] <ProfMikey> yeah that would rock
[02:50:21] * gisburn thinks about throwing ProfMikey into  deep pit filled with komodo dragons, tape the mayhem and post it on MyTube.com to improve his profile ratings.
[02:50:42] <gisburn> Title: "The end of a university prof."
[02:50:55] <gisburn> s/a /an /
[02:54:02] *** stevel has quit IRC
[02:59:23] *** b_e_n_z has quit IRC
[03:01:41] <gisburn> anyone from sun here ?
[03:02:58] <gisburn> dduvall: ping!
[03:03:16] <dduvall> gisburn: pong
[03:03:23] <gisburn> dduvall: who filed http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6480963 ?
[03:03:49] <dduvall> kupfer
[03:03:54] <gisburn> ugh
[03:04:49] <gisburn> dduvall: that's weired because I thought kupfer knows that the Makefile switches adjust libc::wordexp() to use ksh93...
[03:05:58] <dduvall> <shrug>  I haven't looked at the code.
[03:07:24] <gisburn> dduvall: the problem is that libc::wordexp() uses some hidden sun-specific option in Solaris /usr/bin/ksh to do the word expansion, something which no other shell has.
[03:07:51] <dduvall> That's not what the bug is about, though ...
[03:07:59] *** yarihm has joined #OpenSolaris
[03:08:29] <richlowe> for what it it's worth, the CR is correct.
[03:08:56] <richlowe> the very best you can expect of scattered makefile knobs is people will forget about them, and they'll break in ways that don't affect the common case.
[03:09:18] <richlowe> because they *will* end up forgotten and broken, you just hope they fail-safe :)
[03:09:19] <gisburn> richlowe: I would agree but the Makefile switch for libc is unavoidable.
[03:10:27] <alanc> but that's a makefile switch in a single makefile, right?   the bug argues against having to spread them throughout the tree
[03:10:29] <dduvall> Perhaps.  But the bug is only about deliverable objects, not behavioral differences.
[03:11:02] <dduvall> Perhaps Mike was being more specific than he intended, but I don't see that bug as covering the case you're worried about.
[03:11:07] <gisburn> alanc: it is one master makefile fragment which controls four other locations in the tree.
[03:11:31] <gisburn> alanc: e.g. it is one setting, not four
[03:12:04] <gisburn> alanc: changing usr/src/cmd/ksh/Makefile.ksh93switch will do all neccesary adjustments in the whole tree.
[03:12:39] <richlowe> gisburn: I read the CR the same way dduvall did, as an argument against them ending up scattered everywhere.
[03:12:50] <richlowe> uh, as alanc did, sorry.
[03:13:13] <Tpenta> i think th epoint that he is making is that for one product this is not a bad thing. It becomes unmanageable in the future for N where N is a large number and we need to start looking at ways to centralise the customization
[03:13:49] <Tpenta> i dont think that there is an implication that this needs to happen before ksh93 delivery
[03:13:49] <richlowe> Tpenta: I don't agree that it depends on the size of N
[03:13:55] <richlowe> Tpenta: more on the frequency of use, at best.
[03:14:04] <richlowe> the unused ones will be forgotten, break, and then confuse people.
[03:14:08] <richlowe> though not necessarily in that order.
[03:14:14] <Tpenta> i'm just saying that as N grows, it gets unmanageable, so teh better solution is to fix it while N is small
[03:14:27] <richlowe> you also, fairly obviously quickly reach the point where it's infeasible to test them as part of a normal build.
[03:14:35] <richlowe> that does depend on the size of N
[03:14:35] <gisburn> richlowe: I am sure the Makefile.ksh93switch will be used by all OpenSolaris distributions except Solaris from the beginning.
[03:15:17] <richlowe> Tpenta: that I agree with, beyond the fact I think the chance of the switch *itself* breaking exists from the moment the switch integrates.
[03:16:21] <richlowe> gisburn: amount of work to correctly test them increases exponentially with the number of such switches.
[03:18:28] <Tpenta> for myself I'd actually rather se a seperate script/cmd that nightly calls, so other things can use it
[03:18:42] <richlowe> that's what, I think, the CR in question is asking for.
[03:18:52] <Tpenta> for example if you use bldenv and dmake, then yo could run it after the dmake finishes
[03:19:03] <gisburn> richlowe: yes, but the problem is that if libc::wordexp() fails it blows-up SMF and the system will be unbootable.
[03:19:07] <richlowe> not least because it pushes the maintenance burdon onto the people wishing to make such customization.
[03:19:32] <gisburn> richlowe: this is why the switch was added - too many people forgot something and then all hell went loose.
[03:19:41] <Disorganized> happy friday the 13th everyone
[03:19:46] <Tpenta> my read is that he wants the code in nightly
[03:20:02] <richlowe> Tpenta: the hook in nightly.
[03:20:21] <richlowe> Tpenta: such that custom-code runs after 'make install' in his example.
[03:20:26] <richlowe> Tpenta: 2nd paragraph from the end.
[03:20:27] <Tpenta> agreed that hooks need to go into nightly to do the customisation, but it needs to happen outside of nightly
[03:20:44] <Tpenta> ok i see it
[03:20:57] <dduvall> Like the current post_nightly hook, only after the make install part.
[03:21:10] <Tpenta> oh, hello danek
[03:21:30] <dduvall> Hey, Alan.  Gotta run now, though.
[03:21:43] <gisburn> richlowe: doing the switch after make install is in our case.. uhm... unlucky because libc needs to be re-build and some other stuff moved around. that is much more work than doing it as part of make install.
[03:22:06] <Tpenta> so roland, in summary; i don't think that the bug specifically targets ksh93 requiring anything before putback, rather it i sthere to get us thinking about how to do this cleaner before it becomes unmanageable
[03:22:33] <dduvall> Also, like I've said three times so far, the bug only covers filesystem objects, not functionality.  I don't think it covers the wordexp() case.
[03:22:47] <dduvall> Anyway, really running now.
[03:22:55] <Tpenta> that may be worth stating explicitly in teh CR
[03:23:00] <Tpenta> bye danek
[03:24:29] <hile_> howdy alan
[03:24:59] <richlowe> The correct way to deal with wordexp() is pretty obviously to make it not call *any* shell.
[03:26:57] <gisburn> richlowe: which is not that easy than you think since it requires all expansion functionality of a shell.
[03:27:09] <gisburn> richlowe: which ends-up replicating zillions of lines for that.
[03:27:23] <gisburn> richlowe: it would be much easier to use libshell
[03:27:37] <gisburn> richlowe: but that requires to make libshell reentrant first.
[03:28:54] <gisburn> I would already created something like that but right now I am still hacking for our first putback... which was (AGAIN) delayed thanks to the "FCS quality all the time" problem.
[03:29:03] <gisburn> ETA for putback now: Q3/2007
[03:29:30] *** yarihm has quit IRC
[03:29:58] <gisburn> (in the worst case (yes, I am pessimistic and very depressed thanks to these issues))
[03:30:08] *** Seawolf99 has joined #opensolaris
[03:31:34] <Tpenta> oh cool, the next nightly drop will have tamarack
[03:32:13] * gisburn gets even more depressed and looks for a bottle with something like rum...
[03:32:27] <alanc> that's what you get for choosing ON - lots more process and "FCS quality all the time"
[03:32:34] <alanc> other consolidations are looser
[03:32:39] <alanc> (not sure about SFW though)
[03:33:26] * Tpenta runs and hides before the SFW vs ON argument starts again
[03:33:49] <alanc> just look at the JDS discussion about starting a formal code review process soon, since they have none now...
[03:34:09] <richlowe> alanc's just trying to start up recurring arguments? :)
[03:34:12] <alanc> and I've never filled out a WebRTI in my life
[03:34:36] <alanc> just reminding Roland he asked for all of this
[03:34:51] <alanc> he had a choice, and chose the painful way
[03:35:00] <richlowe> and I figure the RTI system is penance.
[03:35:03] <richlowe> so you should start using it :)
[03:35:24] <alanc> you want Xorg 7.2 this decade?
[03:35:27] <alanc> 8-)
[03:36:09] *** ChanServ sets mode: +o Tpenta
[03:37:27] <alanc> besides, I've been up at the beer bash, so I should probably shut up now 8-)
[03:37:44] <Tpenta> you guys get beer bashes?
[03:38:24] <alanc> yeah - every other week
[03:38:27] *** F_roZeN has joined #opensolaris
[03:38:36] <gisburn> Tpenta: sure, the X11 team is even allowed to drink beer during work.
[03:38:37] <richlowe> Tpenta: it's assumed you guys don't need a specific invitation to drink ;)
[03:38:44] <gisburn> Tpenta: look at the X11 code
[03:39:09] <alanc> it's the Solaris/MPK17 beer bash - funded by user contributions,
[03:39:16] <gisburn> Tpenta: and the realise that drinking beer will make you understand even the worst pieces of code.
[03:39:43] <alanc> beer alone cannot explain the cruftier parts of X
[03:39:50] <Tpenta> ahh
[03:39:59] <gisburn> alanc: Wiskey!
[03:40:25] * hile_ doens't even drink
[03:40:55] <gisburn> hile_: you aren't allowed to drink during work. you don't work for the X11 team.
[03:40:58] <alanc> I think there may have been some seriously mind altering substances involved
[03:41:23] <Tpenta> BSD and LSD originated at Berkely. This is probably not co-incidence
[03:41:24] <hile_> well you are in berserkley, alan :)
[03:41:25] <alanc> it's not an X11 thing - the kernel guys were doing it long before we came to the building
[03:41:56] <alanc> it's actually been about a year since the last time I was physically in berkeley
[03:42:47] <gisburn> alanc: I am sure some parts were written under the influence of a "Pan Galactic Gargle Blaster"
[03:44:07] <alanc> Xnews did have a "-zaphod" flag for dual-head mode
[03:44:38] * Tpenta cringes
[03:45:57] <Seawolf99> can someone tell me how to upgrade/install new packages for solaris express from the internet something like apt-get or yum. I am looking for "sudoers"
[03:46:57] <Disorganized> http://img219.imageshack.us/img219/8854/2611867qi8ph2.jpg
[03:47:02] <gisburn> Seawolf99: Try % rm -Rf /* # as user root
[03:47:58] <Seawolf99> gisburn: Thanks, but no thanks, do you have something better
[03:48:12] <alanc> blastwave.org has pkg-get which is similar to apt-get
[03:48:30] <Disorganized> pkg-get rules
[03:48:31] <alanc> nothing included in Solaris out of the box for downloading additional packages
[03:49:13] *** lloy0076 has joined #opensolaris
[03:50:07] <Seawolf99> and how do you upgrade packages for which bugs have been fixed
[03:50:07] *** vmhobbes has joined #OpenSolaris
[03:50:24] <Tpenta> currently upgrade or bfu
[03:50:36] <Tpenta> remember that we are talking development system here
[03:51:06] <Seawolf99> you mean upgrade to a new build
[03:51:15] <Tpenta> yes
[03:51:28] <alanc> yes - Solaris doesn't make individual packages available for updating like Linux distros
[03:51:53] <alanc> the packages are split up the wrong way for that - too many interdependencies to handle
[03:54:15] <lloy0076> I still can't get my Solaris machine to automatically load my Via Rhine card;  in /etc/driver_alias, add_drv has added an alias for "pci1106,3065" which my card is compatible with (according to prtconf -pv). If I ifconfig vfe0 plumb; ifconfig vfe0 dhcp I get a connection (else I wouldn't be here atm) but on boot, it behaves as though the plumb step didn't execute and the dhcpagent says that there's no such device as vfe0.
[03:54:32] *** gisburn has left #opensolaris
[03:54:33] <lloy0076> Hence, I have a working network but not an automatically working network.
[03:55:32] <Seawolf99> so will there be a process to update individual packages, because installing a new build everytime is very time consuming and annoying process
[03:56:28] <hile_> if you're running Solaris (rather than an SXCR build, etc), you could install the patches as they're replaced
[03:57:22] <Seawolf99> yeah that's true, but wouldn't this tedious process scare a lot of developers
[03:57:50] <hile_> i don't see why
[03:58:26] * hile_ even treats his home "prod" machines the same way he'd treat one of work's 6900s
[03:58:35] <hile_> actually better.... they boot boxes willy-nilly
[03:59:45] <Seawolf99> I just installed SX B46 yesterday and already seeing a lot of bugs and also a lot of them fixed, but now you are saying I have to install a new build and that would also mean downloading the entire thing first. that just makes the development process slow
[04:00:21] <alanc> yes - the development process was designed for a team working together with high bandwidth links
[04:01:45] <Seawolf99> well I am also also a high bandwidth link but it looks like this will reduce the developers willing to contribute
[04:01:47] <alanc> you can upgrade some parts individually - it's just those are the large "consolidation" chunks - like you can BFU all of ON, or replace all the X packages, or upgrade your JDS desktop packages as a whole, just not individual pieces like libc or the kernel
[04:04:34] *** cereion has joined #OpenSolaris
[04:05:54] <Error_404> i'm so glad my fileserver's back online
[04:06:07] <Error_404> it was getting pretty annoying not having music
[04:07:03] *** alanc is now known as alanc-away
[04:07:24] <Seawolf99> alanc: thanks, I will try that. I have been using redhat, fedora for a longtime, scratching SX now. Nexenta was cool, but looked like it still needs some work
[04:08:38] *** cereion has left #OpenSolaris
[04:11:29] *** Seawolf99 has left #opensolaris
[04:12:37] *** solaris-user has joined #opensolaris
[04:13:00] <steleman> #ifdef UNDEF_THREADS_HACK #define my_thread_var _my_thread_var #else #define my_thread_var ((my_thread_var())->my_errno #endif
[04:13:33] <steleman> and now for the Piece De Resistance: UNDEF_THREADS_HACK gets defined and undefined recursively depending on the order of #include's and who includes what when
[04:14:02] <steleman> Comrades, this is what i call Quality COde
[04:20:20] *** lloy0076 has left #opensolaris
[04:20:29] <elektronkind> UNDEF_THREADS_HACK
[04:20:35] <elektronkind> that looks familiar
[04:20:38] <elektronkind> mysql ?
[04:20:51] <steleman> uhmmm yeah
[04:22:50] <elektronkind> that's qwailty with a capital w
[04:23:02] <steleman> KWaliTEEEEE
[04:27:23] *** oooo|en_US has joined #opensolaris
[04:27:24] <libkeiser> if the mysql developers were being honest, they just encase every source file inside an #if defined(HACK) ... #endif pair
[04:27:48] <steleman> libkeiser: these aren't #include guards
[04:28:26] <steleman> these are hacks to make the same set of files sometimes thread safe sometimes thread unsafe depending on who compiles them for which library
[04:30:37] <libkeiser> um, i'm well aware of that. i hate to belabor a joke, but i was saying if they were being truthful, ever source file would be wrapped in #ifdef HACK since that what the entirety of mysql is (ok, maybe innodb is slightly better than a hack)
[04:30:55] <steleman> ohhh
[04:30:57] <steleman> d'oh
[04:31:07] <steleman> <---- might be humor challenged this evening
[04:31:25] * libkeiser is apparently too tired to type this evening :(
[04:31:29] <elektronkind> and to get it to even autoconf you would have to do: CPPFLAGS=-DHACK ./configure ...
[04:31:37] <steleman> some companies actually store important data in this
[04:31:46] <elektronkind> you know... sorta rub it in that way
[04:32:13] <steleman> elektronkind: well actually they took care of writing these hacks automagically whether you like them or not :-)
[04:32:42] <steleman> i am seriously beginning to appreciate the beauty and safety of storing important data in simple files
[04:32:46] *** deny_ has joined #opensolaris
[04:32:49] <elektronkind> I need to fire up pg one of these days and get familiar with it.
[04:32:59] <steleman> pg rocks
[04:33:55] <elektronkind> in a world where mysql == OMGWTF and oracle == "dear god 2GB of RAM just to install??!", I'm hoping pg is a sane middle-ground.
[04:34:10] <deny_> i just found out about ZFS and seems to me to be promising, i got to ask though is the feature that identifies a dyin HD earlier part of that package?
[04:34:34] <delewis> memory is cheap :-)
[04:34:57] <elektronkind> deny_: that's not part of zfs. are you thinking of drive SMART monitoring?
[04:36:22] <deny_> i am just downloading bilinx, so i hope it has all those features
[04:37:39] <deny_> i belenix
[04:37:43] <deny_> i mean
[04:38:11] <deny_> what is the version that zfs is on now?
[04:39:16] *** nyati has quit IRC
[04:39:40] <elektronkind> depends who you ask :)
[04:39:58] <elektronkind> the code in opensolaris is ver 2 iirc
[04:40:13] <deny_> i know its not totally finished, i mean you can not make the root folder as zfs format
[04:40:23] <elektronkind> sure you can
[04:40:42] <deny_> or is that the /boot?
[04:41:08] <elektronkind> look at this:
[04:41:09] <elektronkind> http://solaristhings.blogspot.com/2006/06/zfs-root-on-solaris.html
[04:41:28] <elektronkind>  /boot is ufs, though
[04:41:35] <elektronkind> because grub does not understand zfs
[04:41:52] <deny_> ok, thats what i am thinking of
[04:41:54] <elektronkind> perhaps one day it will, but not now
[04:42:31] <deny_> i hope this becomes part of the rest of open source, its a nice file system
[04:42:43] <elektronkind> it already is!
[04:42:45] <Auralis> it already is
[04:42:54] <elektronkind> see the Linux FUSE project and FreeBSD.
[04:43:04] <elektronkind> zfs is being ported there
[04:43:06] <Auralis> and dragonfly bsd and possible osx
[04:43:16] <deny_> i mean, projects like the linux kernel, red hat, suse
[04:43:29] <elektronkind> linux kernel will never have zfs in it
[04:43:37] <elektronkind> license wars and all
[04:43:39] <deny_> y?
[04:44:02] <elektronkind> it boils down to the GPL vs. the CDDL license
[04:44:06] <elektronkind> they are incompatible
[04:44:33] <elektronkind> you can't use CDDL code (such as ZFS) directly in the Linux kernel (which is under the GPL license)
[04:44:40] <deny_> wel gpl v3 is to come out soon
[04:45:04] <elektronkind> and linux won't use the GPL3 license. Linus has already made that pretty clear :)
[04:45:13] <Auralis> that will change nothing, since th kernel is most likely going to stay gpl2
[04:45:17] <deny_> so maybe solaris might like that and then they could meet
[04:45:39] <Auralis> except that gpl2 and gpl3 are incombatible
[04:45:42] <elektronkind> perhaps. there's always a possibility that anything can happen
[04:46:17] <elektronkind> of course, depending on who you ask, the GPL itself is either free or not truly free
[04:46:32] <delewis> or maybe we could just all say "fuck licensing" and get back to actually *using* our systems and writing code.
[04:46:38] <elektronkind> so there's a whole morrass of legal and emotional stupidity involved
[04:47:18] <elektronkind> delewis: that's not your toothbrush, that's OUR toothbrush, comrade.
[04:47:28] <deny_> i got to say, before going further, that i have never used solaris before
[04:47:39] <deny_> and i am looking to try it this weekend
[04:47:46] <elektronkind> cool!
[04:47:54] *** piwi has quit IRC
[04:48:12] <deny_> i just hope it is not too far off from what linux offers
[04:48:20] <elektronkind> belenix, nextenta, SXCR, there are several you can try
[04:48:25] <delewis> deny_: Linux offers little by comparison.
[04:48:53] <delewis> you actually have useful observability tools, resource controls, real volume management, and compliance with standards.
[04:49:07] <deny_> delewis, i don't know about that, they were just saying that there are thousands of project that help linux
[04:50:33] <elektronkind> an open source Linux has also been around for 15 years. OpenSolaris is just past one year :) Projects are forming around it (look at Belenix and Nexenta and so on as examples) which is pretty good being only a year out
[04:51:04] <deny_> by the way, does solaris have a gui? what is it? KDE?
[04:51:05] <elektronkind> it's hard to do a apples to apples comparisson
[04:51:16] <elektronkind> just look at it as a completely new OS and try to explore it deeply
[04:51:17] <delewis> deny_: JDS (a modified Gnome) or CDE
[04:51:32] <Auralis> x11, default is CDE and jds/gnome2, you can run whatever wm you want
[04:51:36] <delewis> but because it's just X11, you're perfectly capable of running whatever window manager/desktop environment on Solaris that you would run on Linux
[04:51:52] <deny_> thanks.
[04:52:48] <deny_> after reading about how solaris is able to detect defects in hard drives before they occur, that was amazing, besides the files system that will handle 2^128 in size, oh boy
[04:53:11] <Auralis> the best about zfs is how easy it is to use
[04:53:19] *** Sieghard has joined #opensolaris
[04:53:26] <delewis> deny_: there are still a lot of features that Solaris has that no other OS has gone before.
[04:53:34] <delewis> DTrace being an excellent example
[04:53:39] <deny_> but it has a md5checksum system built into the filesystem, that was amazing that is so up to date and fresh
[04:53:47] <delewis> well, not MD5
[04:53:53] <delewis> but some sort of cyclic checksum
[04:54:08] <deny_> i thought that the flash sample they say md5
[04:54:34] <LeftWing> MD5 isn't that up to date anyway.
[04:54:51] <deny_> what, there isnt any md6 yet
[04:55:07] <LeftWing> There are other hashing algorithms, though.
[04:55:14] <delewis> fletcher2, fletcher4, and sha256 are the current checksums
[04:55:47] <deny_> so the checksums, oh man i am behind on this
[04:56:25] *** deedaw has joined #opensolaris
[04:57:29] <deny_> thought i did not quite get the c0d0 c1d1 from the samples, that seems to me like the hds themselves
[04:57:30] <elektronkind> sha256 is pretty large
[04:57:50] <elektronkind> they are
[04:58:04] <deny_> actually you would be right about the sha256 because they said the checksum is 256bit
[04:59:07] <deny_> so c0d1 is a hd, what do they really refer to, like c0 is what, does d0 mean disk 0?
[04:59:18] <elektronkind> it needs to be that big since zfs can (theoretically) hold more data than md5 has space for. you'd get collisions with md5 at some point I suppose
[04:59:27] <Auralis> Controler Target Device Slice
[05:00:04] <deny_> auralis, what is that, you are refering to the c0d0 system of naming hds
[05:00:10] <Auralis> yes
[05:00:29] <deny_> you got a way where i could get more docs on this?
[05:00:43] <elektronkind> c0 = controller 0. t0 = target or LUN (in the case of SCSI) number, d0 is drive number (or drive within a SCSI LUN), s0 = slice number
[05:00:47] <Auralis> docs.sun.com  everything you ever want, and then more
[05:01:02] <deny_> ok
[05:01:15] <deny_> thnx
[05:01:47] <elektronkind> so for example, c3t0d0s2 would mean "the device is on controller 3, target (or scsi ID) 0, drive 0, slice 2
[05:02:26] <deny_> slice as you put it, is the actuall plater of the hd?
[05:02:40] <Auralis> slice is a kind of partion
[05:03:04] <deny_> like apple pie slice?
[05:03:15] <elektronkind> yeah, for the sake of argument, "slice" can be thought of as a partition
[05:03:21] <Auralis> more like a logical partion inside a x86 primary
[05:03:52] <deny_> do you mean like the reference to 0x88 location on a hd?
[05:04:01] <elektronkind> sort of like how "sd1a" in linux is the first partition on scsi drive 1, c0t0d0s0 would be the equivalent
[05:04:38] <Tpenta> exept that if you add another drive onto the scsi chain of a linux box, sd1a might change
[05:04:39] <deny_> oh, i got a better understanding now
[05:04:42] <Tpenta> if the scsi id is lower
[05:05:39] <deny_> well i will have to look at docs.sun.com
[05:05:52] <elektronkind> deny_: Linux goes by primary and logical hard drive partitions. So does solaris, but within a solaris-typed primary or logical partition, you have slices.
[05:05:59] <deny_> so i get lots of detail, so i don't bother you all with this
[05:06:09] <Tpenta> and on linux if you add another controller that gets dicovered first electrically, i believe that all bets are off :)
[05:06:37] <deny_> what do you mean all bets are off?
[05:07:18] <elektronkind> so say you have a hard drive with whole thing being one solaris-type primary partition. that would be for example c0t0d0p0. within that primary partition you have slices, and that is what gives you c0t0d0s0-7
[05:07:19] <deny_> you say it now will put ita s sda2 instead of sda1?
[05:07:20] <delewis> Linux's SCSI device is suspect.
[05:07:25] <Tpenta> solaris assigns controller numbers and they stay that way, I believe (well last time i looked), linux will change the sd names if it discovers a controller earlier than the one you had (this may have changed since i last looked)
[05:07:29] <delewis> and scanning re-scanning the SCSI bus is a PITA
[05:07:35] <Auralis> on solaris a device name stays what it is, even if you add new hardware, unless you tell the system to rebuild the device tree from scratch
[05:07:40] <delewis> though, I think that may have changed in later 2.6.x kernels
[05:08:00] <delewis> used to, you had to pipe something to some arbitrary file in /proc to re-scan a SCSI channel
[05:08:24] <delewis> go figure.
[05:08:25] <Tpenta>  /etc/path_to_inst gurantees that the controller numbers in solaris don't change. The target numbers also wont change as the scsi target number is listed as a part of teh device name
[05:08:48] <deny_> awww
[05:09:43] <Auralis> whoch alos comes in very handy in clustering, where each machine has to have the same device names for example, you can use the path_to_inst to make them all the same
[05:09:55] <deny_> well, as i have it so far, my system will have the sda's aranged the same way as long as you do not change the chanel on them
[05:10:44] <deny_> oh , by chance does solaris have its own clustering tools?
[05:11:10] <delewis> define "clustering tools"
[05:11:21] <Auralis> sun cluster, or several thrid party, like veritas, stonebeat and so on
[05:11:28] <elektronkind> deny_: like HPC cluster or HA cluster ?
[05:11:41] <deny_> like in linux you have, heartbeat to make sure that the other system is still alive
[05:11:51] <elektronkind> oh, HA cluster you mean then
[05:11:56] <elektronkind> yes. it's called Sun Cluster
[05:11:58] <delewis> that would be a commonality between HPC cluster and HA clustering tools
[05:12:07] <delewis> well, some HPC clusters, anyhow.
[05:12:12] <elektronkind> some :)
[05:12:28] <delewis> Sun Cluster is free, too :-)
[05:12:50] <deny_> but how does it work, is there like a master or is there a way that they all id themselves and spread the work
[05:13:02] <delewis> "spread the work" is HPC
[05:13:10] <delewis> this is what I meant by defining what type of clustering you mean
[05:13:17] <delewis> it is used *way* too loosely in most cases.
[05:13:22] <deny_> then that is what i mean, i am looking at it as a mosix system
[05:13:36] <delewis> Sun Grid is probably what you want then
[05:13:39] <deny_> or openmosix now
[05:13:45] <delewis> Sun Cluster, Veritas, etc. are all for high-availability
[05:13:49] <deny_> darn, it has lots of tools then
[05:13:50] <elektronkind> deny_: you're talking about compute clusters... yeah, Sun Grid does that.
[05:13:56] <delewis> HPC systems do not guarantee HA
[05:14:08] <delewis> in fact, HPC does not care about HA -- just distributing the workload effectively across all nodes.
[05:14:18] <deny_> that is fine, that is why you would have a nfs storage
[05:15:10] <delewis> NFS isn't a cluster filesystem
[05:15:15] <deny_> i know
[05:15:33] <deny_> but i am refering to it as a storage means for the hpc system
[05:17:38] <delewis> ugh. when did the clock applet default start using mod 12 time instead of mod 24 :-(
[05:17:41] <deny_> well i am thinking of building a "ring" system, where the pc would be doing the calculations and then have nfs attached and such
[05:18:19] <deny_> so if someone attacks the firewall i could increase the computing power and defense power to that
[05:18:20] <elektronkind> I'd check out sun grid (aka N1 Grid System)
[05:18:34] <elektronkind> ?
[05:18:38] <delewis> which is also completely free.
[05:18:53] <delewis> deny_: sounds like something from Star Trek
[05:18:59] <delewis> "more power to the shields!"
[05:19:14] <elektronkind> rotate the shield harmonics
[05:19:14] <deny_> actually it does, since you put it that way
[05:19:16] <elektronkind> (.pl)
[05:19:19] <delewis> hehe
[05:19:29] <delewis> "oh noes, they know our shield frequencies!"
[05:19:32] <delewis> yeah.
[05:20:04] *** laca has quit IRC
[05:20:05] <Error_404> sun apparantly is in the consulting business now, since it appears they don't make software
[05:20:17] <delewis> well, support, anyhow.
[05:20:27] <deny_> but that is what would be nice to you a hpc for, at least at full capabilities
[05:22:27] <deny_> *use
[05:22:53] <elektronkind> a HPC cluster would be a nice thing to have right now to help me figure out what I should make for dinner.
[05:27:25] <deny_> lol
[05:27:54] <deny_> pizza is always good, if you got ingrediences
[05:29:06] <LeftWing> Or a phone.
[05:29:17] <Error_404> heh
[05:29:27] <delewis> well, right now, I'm trying to figure out whether or not the cash for the Acer Ferrari 5000 is worth it over the 4000 series.
[05:29:42] <deny_> price?
[05:29:51] <deny_> and specs
[05:31:10] <delewis> http://us.acer.com/acereuro/page4.do?sp=page3&dau22.oid=10478&UserCtxParam=0&GroupCtxParam=0&dctx1=25&CountryISOCtxParam=US&LanguageISOCtxParam=en&ctx3=-1&ctx4=United+States&crc=123596151
[05:31:27] <delewis> http://us.acer.com/acereuro/page4.do?sp=page4&dau22.oid=18364&UserCtxParam=0&GroupCtxParam=0&dctx1=25&CountryISOCtxParam=US&LanguageISOCtxParam=en&ctx3=-1&ctx4=United+States&crc=2670708976
[05:32:06] <delewis> Froogle indicates about $1.8k USD for the 4000 and about $2.2k USD for the 5000.
[05:32:31] <deny_> u looking at it fogaming/
[05:32:42] <delewis> deny_: portable solution for development
[05:32:59] <delewis> it's not always practical to connect to my SB1000 in my home office from wherever I happen to be
[05:33:09] <Error_404> heh... gaming
[05:33:20] <delewis> (grab the SSH keys, install Java, and connect to my SGD server)
[05:33:33] <deny_> putty
[05:33:40] <delewis> it'd be much easier to just whip out a Ferrari with Solaris installed :-)
[05:33:52] <delewis> deny_: and what happens when I need X11? :-)
[05:33:59] <deny_> vnc
[05:34:14] <delewis> that's what SGD accomplishes.
[05:34:16] *** dclarke has joined #opensolaris
[05:34:21] <delewis> with much more finesse
[05:34:26] <delewis> VNC is a hack by comparison
[05:34:31] * dclarke falls in the door
[05:34:38] <deny_> alright u need the 4000 ferrari
[05:34:43] * LeftWing fishes dclarke out of the door.
[05:34:45] <delewis> welcome, dclarke
[05:34:55] <Error_404> good evening drunkclarke
[05:34:58] <dclarke> g'day
[05:34:59] <deny_> but idont see why u cant get ahp per say
[05:35:07] <delewis> well, one of the criteria I'm also placing on a portable solution is 64-bit
[05:35:16] <delewis> Core 2 Duo laptops are still far and few
[05:35:23] <dclarke> Error_404:  I got your ODW in a box
[05:35:26] <delewis> and Acer seems be one of the few companies pushing AMD64
[05:35:27] <dclarke> Error_404:  finally
[05:35:28] <deny_> turino amd
[05:35:39] <Error_404> oh, rad
[05:35:41] <dclarke> Error_404:  and then found a use for it .. but I'll ship it anyways
[05:35:47] <Error_404> heh
[05:35:54] <Error_404> did that intellistation not pan out?
[05:36:07] <dclarke> Error_404:  the bit PowerPC op ?
[05:36:15] <dclarke> Error_404:  the big PowerPC unit ?
[05:36:28] <dclarke> Error_404:  they bailed out for "political" reasons
[05:36:33] <delewis> Intellistations == p6xxs (usually p610s)
[05:36:33] <Error_404> the one that IBM gave you & didn't want anyone to know
[05:37:29] <dclarke> IBM may still arrive at the door also
[05:37:32] <delewis> interesting
[05:37:35] <dclarke> I'll just keep my mouth shut and hope for the best
[05:37:39] <delewis> never thought IBM would pony up some hardware for OpenSolaris development
[05:37:44] <delewis> though, if they do, I'll be pleased.
[05:37:51] * delewis would like to see Solaris on his p640 :-(
[05:38:01] <deny_> what u want, the cell system?
[05:38:08] <Error_404> delewis: perhaps they noticed solaris' growth?
[05:38:18] <delewis> Error_404: and the demise of AIX, probably.
[05:38:36] <delewis> I like AIX, but it's *really* lagging in terms of features that Solaris has.
[05:38:38] <dclarke> I was actually talking with the BlueGene people
[05:38:45] <dclarke> they had a bit of a fit at the idea
[05:38:46] <delewis> similar situation with IRIX years ago
[05:39:01] <delewis> extraordinary hardware, average UNIX, but nothing unique.
[05:40:12] <Error_404> dclarke: they were gonna let you install solaris on bluegene?
[05:40:19] <dclarke> no way
[05:40:35] <Error_404> yeah, that'd be wierd
[05:40:40] <dclarke> I suggested that we could look into working on the OpenSolaris code with their tools
[05:40:47] <dclarke> they said no way to that too
[05:40:54] <deny_> why?
[05:41:02] <delewis> dclarke: what tools do they have?
[05:41:06] <dclarke> for obvious bloody reasosn
[05:41:16] <deny_> why they say no
[05:41:26] <delewis> they'd be betraying their AIX and Linux customers :-)
[05:41:27] <dclarke> excelllent compilers and tools for PowerPC hardware
[05:41:34] <dclarke> bingo
[05:41:54] <delewis> dclarke: tool-wise xlc is pretty lacking, and gcc generates fairly decent code for PowerPC, nowadays.
[05:42:28] <delewis> though, some I hear some good things about Altivec and it, unfortunately, I haven't had a chance to play around with any of that.
[05:42:39] <deny_> "wouldn't you like to have that gibson, baby" HACKERS
[05:42:50] * delewis POWER lacks Altivec, and AIX Altivec support is very, very new
[05:42:53] <Error_404> i must say, i don't follow how it would be a betrayal to their other customers, but *shrug*
[05:43:08] <dclarke> its of no concern
[05:43:23] <dclarke> at the moment I am looking into UltraSparc assembly
[05:43:30] <delewis> Error_404: AIX customers have been guaranteed support and if IBM starts putting some of their eggs in another basket (OpenSolaris), it'll appear they're giving up on AIX.
[05:43:32] <dclarke> trying to get a handle on VIS instructions
[05:43:57] <delewis> and I could imagine those customers would be extremely observant, given the HP fiasco with VMS and Tru65.
[05:44:00] <delewis> Tru64*
[05:44:19] <Error_404> odd, it's just diversifying
[05:44:20] <dclarke> be right back ...
[05:44:24] <Error_404> but then again, i'm not corporate-guy
[05:44:59] <delewis> Error_404: how would you feel if you just furnished your datacenter with VMS hardware and HP announces they are working on something else and stopping VMS development?
[05:45:06] *** kloczek has quit IRC
[05:45:18] <LeftWing> delewis: That I was foolish enough to choose HP? ;P
[05:45:25] <delewis> well :-)
[05:45:30] <delewis> that's besides the point.
[05:45:49] <LeftWing> http://h71028.www7.hp.com/ERC/cache/107844-0-0-0-121.html?ERL=true
[05:45:57] <LeftWing> You should be The Real Story. =P
[05:46:02] <LeftWing> *get
[05:46:21] <delewis> well, if my analogy doesn't apply to VMS, then it applies to HP-UX
[05:46:41] <delewis> (though, IIRC, HP released an update about a year ago -- HP-UX had gone without updates for some time)
[05:47:14] <LeftWing> In August 2005, HP delivered security containment for HP-UX 11i version 2 secure resource partitions.  Now, HP-UX 11i delivers what other UNIX vendors don?t even claim - mission critical virtualization. And, it works without changes to your existing application and database software.
[05:47:27] *** dclarke has quit IRC
[05:47:29] <Error_404> i'd feel like what i bought serves the purposes that i bought it for, so whatever
[05:47:32] <Error_404> but that's me
[05:47:37] <delewis> I would certainly hope they would preserve compatibility between updates!
[05:47:57] *** deny_ has quit IRC
[05:48:09] <delewis> Error_404: when corporations choose a vendor and implement that vendor's solutions in their datacenter, it's for the long-term.
[05:48:30] <LeftWing> delewis: It's the "what other UNIX vendors don't even claim" part that irks me.  It's blatantly false.
[05:48:31] <delewis> knowing your vendor is working on something else tends to taint the long-term viability of that solution you implemented.
[05:48:44] <delewis> LeftWing: it's obvious they're advertising to Linux/Microsoft shops.
[05:48:53] *** quellhorst has joined #opensolaris
[05:49:08] <LeftWing> delewis: Not really, if you look at the rest of the "Real Story" articles they're bashing Sun and IBM.
[05:49:27] <delewis> LeftWing: they're just taking advantage of the FUD from the Solaris 7 and 8 days
[05:49:36] *** deny_ has joined #opensolaris
[05:49:37] <delewis> "Migrate to Linux!"
[05:49:46] <delewis> "Away from your ancient enterprise UNIX"
[05:49:48] <LeftWing> It's unfortunate that managers are all about the FUD. =P
[05:50:16] <deny_> FUD sells
[05:50:21] <delewis> unfortunately
[05:50:30] <LeftWing> Our idiot manager tried to rib us that "it must be hard for you Solaris admins to swallow your Solaris running on Windows!"  (he was talking about VMware ESX)
[05:50:44] <delewis> which is why I'd like to stay out of the IT industry.
[05:50:59] <delewis> experience is irrelevant. knowledge is irrelevant. FUD is relevant. "the next biggest thing" is relevant.
[05:51:05] <deny_> how about you become the manager?
[05:51:23] <LeftWing> Become the manager?
[05:51:39] <delewis> VMware ESX isn't Windows :-)
[05:51:42] <deny_> be a manager, and say bye to FUD
[05:51:55] *** solaris-user has quit IRC
[05:51:57] <delewis> it's a modified Red Hat 7.2, IIRC.
[05:52:02] <delewis> at least ESX 2.5 was
[05:52:09] <quellhorst> so, after i boot the dvd. i selected the first book option. then i get a prompt 1-6 to choose starting the interactive, jupmstart, desktop session, console session. drive updates. single user shell...
[05:52:12] <LeftWing> delewis: I'm aware of that.  He just hates Solaris and Solaris admins, seemingly, and has no actual technical or even operational knowledge.
[05:52:15] <quellhorst> how do i install opensolaris?
[05:52:22] <delewis> LeftWing: imagine that.
[05:52:23] <LeftWing> Which is good because we're mostly a Solaris on SPARC shop.
[05:52:30] <delewis> IT managers with experience.
[05:52:32] <delewis> who would've thought?
[05:53:13] <LeftWing> delewis: We lost our last available competent UNIX admin to resignation the other day.  It looks like the manager is going to try and hire a Windows admin to replace him.
[05:53:30] <delewis> the academic/research industry is looking better and better compared to IT.
[05:53:49] <LeftWing> This is academia. I'm talking about a University. ;)
[05:54:02] <delewis> well, University IT departments are always suspect. :-)
[05:54:23] <LeftWing> haha
[05:54:32] <quellhorst> i got tho sol-nv-b49-x86-dvd-iso files.. is this the incorrect dvd to install opensolaris?
[05:54:52] <LeftWing> Our department used to be much less suspect, but this manager is making everybody want to quit.  And this is coming after most of the good staff were made redundant in a large cost-cutting measure earlier.
[05:55:03] <delewis> quellhorst: no, that DVD is what you want.
[05:55:21] <quellhorst> ok, then im at a loss on how to install from it.
[05:55:34] <delewis> boot the CD and choose Interactive Install
[05:56:00] <quellhorst> it doesnt say install, it just says `1 solaris interactive (default)`
[05:56:09] <delewis> that's what you want.
[05:56:10] <quellhorst> if i choose that then it complains about no X11
[05:56:34] <LeftWing> Interactive console, then?
[05:56:46] <quellhorst> 'ERROR: cannot find windowing software, exiting to shell'
[05:56:54] <LeftWing> .. Cannot find it?
[05:57:05] <quellhorst> thats option one
[05:57:09] <delewis> quellhorst: did you concatenate the files correctly and did you verify the checksums?
[05:57:24] <quellhorst> i did concat it correctly
[05:57:33] <quellhorst> will have to check the checksum
[05:57:48] * quellhorst hits himself... deleted the .zips
[05:57:57] <delewis> oh, I forget
[05:58:00] <delewis> .zips have internal checksums
[05:58:04] *** nachox has joined #opensolaris
[05:58:15] <delewis> if the file was bad, when you unzipped, the zip unarchiver would've caught it.
[05:58:22] <quellhorst> "cannot find install software, exiting to shell"
[06:00:47] <elektronkind> maybe it went out to 7-11 for a coffee and will be back soon
[06:00:57] <elektronkind> maybe that's what I need to do
[06:01:02] * elektronkind &
[06:01:22] <quellhorst> im gonna try downloading all the files again ;(
[06:02:53] <nachox> elektronkind: ctrl-z :)
[06:04:45] *** lloy0076 has joined #opensolaris
[06:04:48] *** Sieghard has quit IRC
[06:07:32] <boyd> Hmm.... is there any way for me to add apps to the "Applications" menu in JDS?
[06:07:38] *** Hellaenergy has quit IRC
[06:08:02] <boyd> Actually, make that "All Applications->Developer Tools"
[06:08:36] *** Sieghard has joined #opensolaris
[06:08:41] *** kloczek has joined #opensolaris
[06:09:36] *** lloy0076 has quit IRC
[06:10:29] <delewis> boyd: I've wondered about that, too
[06:10:39] <delewis> unfortunately, it seems all of the menu configuration files are owned by root
[06:10:44] <delewis> and are thus non-local
[06:11:19] <boyd> Umm... That seems... lame.
[06:11:22] <delewis> it does
[06:11:32] <delewis> it also means you can't even edit entries in the menu
[06:11:33] <boyd> I thought it might read a user-specific set of files too.
[06:11:46] <nachox> kde rocks!!... err... :)
[06:11:57] *** jcsmith has quit IRC
[06:12:01] <delewis> boyd: perhaps, but I don't know what the user-specific ones are
[06:12:18] <boyd> You can disable the ones that are there, but it seems not add your own. Since you can make your own kicker things it seems inconsistent
[06:12:23] <delewis> but if you try editting, say the "Firefox" entry, you'll see that you can't
[06:12:40] <delewis> if there were user-specific ones, that should be possible, I would think
[06:12:46] <boyd> Where's Gman when you need him?
[06:13:17] *** solaris-user12 has joined #opensolaris
[06:13:39] <boyd> And another thing, I want to be able to split a gnome-terminal tab from the window it's in.
[06:13:56] <boyd> Oh, wait..
[06:14:10] <boyd> Good grief, I can :)
[06:14:31] <boyd> ... althought I can't seem to undo it.
[06:15:17] <Tpenta> try dragging it back in; it works in gaim
[06:15:41] <boyd> Tpenta: It's not apparent what I drag it in *to*
[06:15:54] <Tpenta> back to the tabs
[06:16:11] <Tpenta> or perhaps teh title bar
[06:16:30] <boyd> Hmm... seems not
[06:17:23] *** Teknix has quit IRC
[06:17:54] <richlowe> bugger.
[06:18:23] *** jp_n9 has joined #opensolaris
[06:20:59] *** nyati has joined #opensolaris
[06:23:33] *** ShadowHntr has quit IRC
[06:26:33] *** jp_n9 has left #opensolaris
[06:29:27] *** Teknix has joined #opensolaris
[06:33:16] * boyd plays with Chime... it's a little better than he expected.
[06:39:39] <jbalint> whats that?
[06:40:24] <nachox> http://www.opensolaris.org/os/project/dtrace-chime/install/;jsessionid=4270FBE6066E092363E577EAD3D3ACAF
[06:41:24] <jbalint> Ah, neat.
[06:41:51] <elektronkind> can you plug your own scripts into it?
[06:42:33] <nachox> parently they have to be adapted but yes
[06:57:33] *** nachox has quit IRC
[06:57:59] *** solaris-user12 has quit IRC
[06:58:34] <boyd> Actually, many of them need little adaption, but they need to be wrapped in an XML file. There is a wizard to do that fortunately.
[06:59:17] * boyd wonders how much of a security hole a w.o.o session id from someone else is :)
[07:12:32] *** F_roZeN has quit IRC
[07:14:13] *** Kernel86|Laptop has joined #OpenSolaris
[07:28:21] *** Sieghard has quit IRC
[07:29:18] *** Sieghard has joined #opensolaris
[07:36:04] *** gtc has joined #opensolaris
[07:51:19] *** vmhobbes has quit IRC
[07:59:46] *** vmhobbes has joined #OpenSolaris
[08:32:23] <twincest> does anyone support connecting to AF_UNIX sockets over nfs?
[08:34:37] <Sieghard> cannot believe my pool just died on me
[08:34:57] <Sieghard> don't reset your machine without exporting your pool kids
[08:44:38] <Error_404> define 'reset' ?
[08:50:57] *** lloy0076 has joined #opensolaris
[08:54:10] <jmcp> Sieghard: don't export your root pool if you want to be able to boot from it
[08:54:29] <Sieghard> Error_404: press reset button
[08:54:49] <jmcp> yeah, that would be a less-than-sensible thing to do
[08:55:06] <Sieghard> well the machine froze
[08:55:07] <twincest> zfs is meant to recover from that with no issues
[08:55:24] <jmcp> Sieghard: x86/x64 box?
[08:55:25] <Sieghard> twincest: i hoped too, but it came back with a faulted pool
[08:55:29] <twincest> (well expect some lost data perhaps, butit shouldn't corrupt the pool)
[08:55:30] <Sieghard> jmcp: x64
[08:55:43] <twincest> sieg: s10 or sx?
[08:55:48] <Sieghard> twincest: it says "metadata corrupted"
[08:55:55] <Sieghard> twincest: sol10u2
[08:55:57] <jmcp> Sieghard: that's why I always boot my systems with -k, so I can try the old F1-A trick
[08:55:59] <jmcp> ouch
[08:56:16] <Sieghard> F1-A trick?
[08:56:35] <jmcp> yeah, booting with -k allows you to fall through to kmdb
[08:56:42] <jmcp> by hitting F1 and A
[08:56:57] <twincest> F1 is like stop on x86?
[08:57:00] <jmcp> then you type                    0>rip [enter] :c [enter] :c [enter]
[08:57:01] <jmcp> yeah
[08:57:06] <Sieghard> oh, well i wasn't even expecting the machine to hang
[08:57:14] <jmcp> and 0>eip if you're on x86 (rip is x64)
[08:57:17] <Sieghard> it just hung, and both ethernet links where down
[08:57:28] <Sieghard> *were
[08:57:35] <jmcp> Sieghard: most people don't expect their machines to hang
[08:57:36] <twincest> oh nice, secure-by-default is getting retrofitted to S10
[08:58:33] <Sieghard> i want the don't-crash-by-default feature ;)
[08:58:47] *** vmhobbes has quit IRC
[08:59:28] <richlowe> jmcp: $<systemdump
[09:00:04] <jmcp> richlowe: that too
[09:03:09] <lloy0076> ld.so.1 will look at what the command "crle" says is the default regardless of what I have in /var/ld/ld.config won't it?
[09:03:20] <Error_404> Sieghard: tried scrubbing the pool?
[09:03:55] <Sieghard> Sieghard: you can't scrub a faulted pool, it won't mount
[09:04:12] <Sieghard> Error_404: err.. the above was supposed to be to you
[09:04:17] <richlowe> Why does zpool status -x say the pool is faulted.
[09:04:27] <Error_404> hey wtf ?
[09:04:36] <Error_404> i have half a bottle of rum in the freezer
[09:04:40] *** slowhog has left #opensolaris
[09:04:45] <Error_404> between some fish sticks & peas
[09:04:46] <richlowe> Error_404: the *freezer*?
[09:04:46] <jmcp> Error_404: that's the wrong place for it
[09:04:59] <Error_404> i don't think i put it there tbh
[09:05:10] <Sieghard> maybe it's so no one will find it
[09:05:21] <Sieghard> and think it's maple syrup or something
[09:05:39] <Error_404> maple syrup doesn't slosh around like rum
[09:05:42] <Sieghard> richlowe: metadata is corrupted
[09:05:45] <Error_404> and it usually doesn't smell like booze either
[09:06:17] <Sieghard> Error_404: I thought it'd be frozen, but i guess the freezer is not cold enough
[09:06:39] <Error_404> it's hard liquor... it doesn't freeze that easily
[09:07:18] <Sieghard> Error_404: is it like 180 proof
[09:12:05] <Error_404> 40% alcohol/vol
[09:12:48] <Sieghard> Error_404: oh, as strong then, should freeze
[09:12:53] <Sieghard> i mean it's not as strong
[09:13:14] <Error_404> *shrug*
[09:13:47] <lloy0076> Error_404: Drink the whole lot now just in case it *does* freeze...
[09:14:10] <Error_404> yeah, that's what i was thinking
[09:14:35] <Sieghard> microwave it first
[09:14:42] <Error_404> whoever put it there is going to have a suprise next time they're depressed
[09:14:57] <Error_404> Sieghard: why on earth would i microwave it?
[09:15:02] <Sieghard> Error_404: doesn't drinking make you even more depressed?
[09:15:17] <Error_404> Sieghard: yeah, but it also makes you not care
[09:15:27] <Sieghard> i guess
[09:16:23] * lloy0076 sigh
[09:17:04] <lloy0076> Seeing that libreadline.so is in /usr/local/lib and I put --libdir="/usr/local/lib" in postgresql-8.1's configure, configure decides not to pass the include directive to gcc.
[09:17:17] <Sieghard> strange
[09:17:39] <lloy0076> So, postgres is going to be built without readline support :)
[09:18:06] *** gtc_ has joined #opensolaris
[09:24:37] *** krozinov has quit IRC
[09:28:40] <lloy0076> Is there an official Adobe Acrobat for Solaris x86? I can't see one from the main download the reader page...
[09:28:53] *** Griffous has quit IRC
[09:30:06] <tsoome> lloy0076: no...
[09:30:34] <Error_404> xpdf ?
[09:30:49] <lloy0076> I've got a viewer.
[09:31:20] <Sieghard> oh, you want the Adobe Acrobat
[09:31:23] <Sieghard> not the reader
[09:34:20] <lloy0076> I've downloaded Open Office 2.0.4 Solaris i86 packages, made an admin file which stops pkgadd from checking dependencies, conflicts and actions and then done "pkgadd -a admin -d . openoffice.org*" only to get "No package associated with openofficeorg-base".
[09:35:00] *** gtc has quit IRC
[09:35:01] <lloy0076> i.e. I seem to have a series of directories which seem to have some form of common structure - presumably solaris packages - and pkgadd isn't recognising them OR I'm doing something wrong.
[09:38:03] <Sieghard> lloy0076: does it work without the admin file?
[09:39:07] <twincest> "SDN"?  this isnew
[09:42:07] <lloy0076> Sieghard: Now.
[09:42:40] <Sieghard> lloy0076: now
[09:43:22] <lloy0076> Sieghard: Sorry, I meant no.
[09:43:29] <lloy0076> Sieghard: The "w" was a Freudian slip.
[09:44:06] <Sieghard> lloy0076: not sure then what could be wrong, have you tried running the openofficeorg-base package on its own?
[09:45:13] *** Teltariat has joined #opensolaris
[09:47:46] <lloy0076> Sieghard: I wouldn't know how to - the "package directories" look like unzipped/gunzipped (untarred?) packages.
[09:48:18] <Sieghard> lloy0076: instead of openoffice.org* just do the base directory
[09:50:06] <lloy0076> Sieghard: That seemed to do it.
[09:50:09] *** Tpenta has quit IRC
[09:50:21] <lloy0076> Sieghard: Now I think I'll write a little bash script and set it off :)
[09:51:09] *** LordKing has joined #opensolaris
[09:53:44] *** gtc_ has left #opensolaris
[09:55:28] <Sieghard> lloy0076: don't think it should be necessary, but yeah
[09:55:41] <lloy0076> Sieghard: I think I found the problem.
[09:56:00] <lloy0076> Sieghard: The first one, base, wanted a response from the command line, spat something out to STDOUT and got pkgadd confused.
[09:56:52] <Sieghard> lloy0076: probably not made for unattended installs?  sad
[10:01:45] *** loke has joined #opensolaris
[10:05:37] *** Yamaraj has joined #opensolaris
[10:09:19] *** Tpenta has joined #opensolaris
[10:10:06] *** alanc_away has quit IRC
[10:10:51] *** alanc_away has joined #opensolaris
[10:27:05] *** vmhobbes has joined #OpenSolaris
[10:27:58] *** Cybernd has joined #opensolaris
[10:35:05] <quellhorst> so i have the sol-nv-b49-x86-dvd.iso image that i am booting off of. when i try the #1 option for solaris interactive it says "error: cannot find wondowing software, exiting to shell"
[10:35:18] <quellhorst> i even redownloaded everything and reverified the md5 sum is correct.
[10:42:25] *** Fish has joined #opensolaris
[10:43:11] <Fish> hello
[10:45:16] <Error_404> ahoy!
[10:47:56] <sickness> quellhorst: so try 4
[10:48:14] <quellhorst> sickness: diff error there. like no install image found
[10:48:40] <lloy0076> The Gimp is getting to be slower than Photoshop to load.
[10:50:12] <Stric> at the first run, it executes all plugins.. second+ time, it's cached..
[10:51:30] <lloy0076> Fair enough.
[10:52:22] <sickness> quellhorst: no install image found? are you trying inside qemu? you have to disable the atapi cd dma mode
[10:53:07] <quellhorst> whats qemu?
[10:54:01] <Yamaraj> An emulator.
[10:54:43] <quellhorst> no, not qemu, but parallels
[10:55:31] <Error_404> oh, bingo
[10:55:59] <Error_404> so it's not such a wierd error after all
[10:57:23] *** ChanServ sets mode: +o Tpenta
[11:01:12] <sickness> yeah
[11:01:14] <quellhorst> cool, looks like thats working
[11:01:27] <sickness> try to disable all the cdrom speed enhancements if you can in the options
[11:01:30] <asyd> ola
[11:01:33] <sickness> like atapi dma :)
[11:01:35] <sickness> cool :)
[11:01:46] <asyd> 30 minutes before my talk about PKI pff, the stress comes
[11:01:55] <sickness> now off to the gym :)
[11:02:39] *** LordKing has quit IRC
[11:02:56] *** bunker has joined #opensolaris
[11:03:24] <Error_404> asyd: drink?
[11:03:49] <asyd> :)
[11:03:58] <asyd> well, I was a bit drunked yesterday :)
[11:04:39] <quellhorst> how can i configure dvorak on solaris?
[11:12:47] <Ludo_> setxkbmap ?
[11:12:56] <Ludo_> loadkey ?
[11:14:39] <quellhorst> man, im kinda shocked how macs are these days.. im on a 3 display box(including 1 30")), and quad cores (xeon) with virtualization accelleration.
[11:14:47] *** Yamaraj has quit IRC
[11:18:24] *** Xh4 has quit IRC
[11:23:16] *** vmhobbes has quit IRC
[11:32:23] *** vmhobbes has joined #OpenSolaris
[11:34:18] *** lloy0076 has left #opensolaris
[12:05:12] <quellhorst> man, this is wild running solaris under an emulator
[12:08:07] *** Disorganized has quit IRC
[12:16:58] *** vmhobbes has quit IRC
[12:19:45] <ProfMikey> quellhorst: yeah man... I am about to get the new macbook 13", its amazing
[12:20:04] <quellhorst> thats a good laptop
[12:20:32] <ProfMikey> and pretty portable
[12:20:50] <quellhorst> yeah, only thing i can say bad about it is that you cant hook up a 30" to it
[12:20:53] *** Kernel86 has quit IRC
[12:20:57] <quellhorst> but it does a 24" fine
[12:21:50] <quellhorst> but thats probably not on the requirements list for most people. i got one with the min specs then ordered 2gig ram and 120gb sata drives from newegg
[12:21:53] <ProfMikey> have 24" dell widescreen
[12:22:03] <quellhorst> you will be fine then
[12:22:12] <ProfMikey> fine :)
[12:23:48] <ProfMikey> yeah the one with 2 gig ram is pretty cool anyway the idea is to get the one with 1 gig and get one more stic elsewhere cuz apple's prices on ram are the killing ones
[12:25:11] <ProfMikey> its like 250 quid difference between 1gb and 2gb models
[12:25:12] <quellhorst> i got the min spec model
[12:25:28] <quellhorst> does the 1gb model have 2x512 or 1x1024?
[12:25:48] <ProfMikey> awww from the newegg, sorry pal my ignorance heh
[12:26:00] <ProfMikey> 1x1024 I guess
[12:26:09] *** Gr|ffous has joined #opensolaris
[12:26:21] <quellhorst> my performance was garbage until i took out the 512... which was 2 x 256
[12:27:00] <ProfMikey> anyway can put it on the ebay if this is 2x256
[12:27:27] <quellhorst> yeah. and the 120gb drive was maybe $150
[12:27:44] <ProfMikey> I dont need so big drive
[12:28:00] <quellhorst> i was messing with some huge datasets.
[12:28:15] <ProfMikey> I am not bacupin all the pron of the net....yet :)
[12:28:25] <quellhorst> haha
[12:28:37] <quellhorst> thats why im looking into solaris... ZFS ;)
[12:29:54] <ProfMikey> "SIR, we cant hire you...you have the exprience, skills we need and all... but we are kinda woried about your blog, you still write about pron...we are afraid you'll use OUR disckspace to store em"
[12:30:30] *** Symmetria has quit IRC
[12:30:50] <ProfMikey> 1gb on newegg is like $130
[12:34:06] <quellhorst> yeah, i got 2 of those
[12:34:29] <quellhorst> was a few months back. not sure if they are still the cheapest
[12:35:02] *** bunker has quit IRC
[12:35:02] <ProfMikey> there are others for 99, unbuffered
[12:35:14] <quellhorst> sure they are the right size?
[12:36:28] <ProfMikey> well pretty much, it is written on the sidebar that the item is compatibile with macbooks
[12:36:56] <quellhorst> nice
[12:37:40] <quellhorst> as for porn... all the datacenters sure do ask me what im doing when i ask about space.
[12:37:51] <quellhorst> especially when i tell them file hosting and ask for 100mpbs
[12:42:05] <ProfMikey> haha
[12:51:36] *** koolniczka has joined #opensolaris
[12:51:54] <sickness> i'm back
[12:58:29] <sickness> did I read pr0n? =)
[12:58:30] <sickness> ghgh
[13:08:04] *** n0rus has joined #opensolaris
[13:20:44] *** Ludo_ has quit IRC
[13:29:00] *** bunker has joined #opensolaris
[13:38:05] *** Kernel86 has joined #OpenSolaris
[13:41:30] *** Kernel86 has quit IRC
[13:41:38] *** lloy0076 has joined #opensolaris
[13:49:20] *** Kernel86 has joined #OpenSolaris
[13:50:18] *** dunc has quit IRC
[13:51:05] *** dunc has joined #opensolaris
[13:51:36] *** deedaw has quit IRC
[13:59:33] *** splitscreen has joined #opensolaris
[14:04:35] <twincest> what can cause an SMF server to be in the uninitialized state with no nstate?
[14:13:24] <twincest> ah, inetd not running
[14:18:05] *** ThEbUtChE has joined #opensolaris
[14:51:46] *** jsubl2 has joined #opensolaris
[14:56:06] <dunc> twincest :) that's an amusing nick
[15:00:00] *** kimc has joined #opensolaris
[15:01:33] <lloy0076> Amazing..
[15:01:49] <lloy0076> Subversion finally compiles without having uber-errors.
[15:02:58] <lloy0076> <-- spoke too soon
[15:03:18] *** n0rus has quit IRC
[15:05:45] *** bunker has quit IRC
[15:07:53] *** bunker has joined #opensolaris
[15:08:38] <kimc> I've recently built subversion on Opensolaris, whats the problem ?
[15:09:15] <kimc> please describe the problem
[15:09:50] <bunker> anyone know how to display system info under solaris from cmdline? (except dmesg)
[15:10:09] <movement> bunker: what sort of system info?
[15:10:26] <bunker> movement: cpu, cache, memory and so on
[15:10:57] *** Mdx4 has joined #opensolaris
[15:10:58] <bunker> found! /usr/platform/`uname -i`/sbin/prtdiag -v
[15:11:28] * bunker thanks google :)
[15:13:48] <lloy0076> Oh, sorry.
[15:14:07] <lloy0076> I've managed to get it built with https support by jumping through a number of hoops.
[15:14:34] <movement> bunker: don't use that. usr /usr/sbin/prtdiag -v
[15:14:40] <lloy0076> Eventually, I managed to stack on enough --blah-blah options to get it to actually compile and not actually fail on a compile/library error.
[15:14:55] <bunker> movement: yes, thanks
[15:15:10] <lloy0076> Then I got a "blah cannot be install in a directory not ending in /usr/sfw/lib" error and I've done a ./configure with yet another --blah-blah option.
[15:15:49] <lloy0076> Those on google seem to think that my latest error is caused by executing two ./configures both with a differing --prefix option.
[15:17:24] *** azr|el has quit IRC
[15:26:04] <lloy0076> kimc: Well, everything installed except for the man pages.
[15:28:21] *** lloy0076 has left #opensolaris
[15:47:11] <kimc> lloy0076: thats great.. you've got it :)
[15:47:50] <kimc> now see what happens if you try to get the opensolaris bits via svn
[16:13:27] *** aramdune has joined #opensolaris
[16:14:07] <aramdune> hello. for installing (compiling) OpenSolaris I need Solaris Express. Can't I compile it in Solaris 10 or Nexenta? Thanks.
[16:16:21] <oxygene> nexenta might work
[16:17:35] <aramdune> but what is specific cu Solaris Express and needed for build that Solaris 10 doesn't have?
[16:17:43] <aramdune> s/cu/to
[16:20:14] *** nyati_ has joined #opensolaris
[16:30:45] <jteo> aramdune: it's in the readme.
[16:31:05] <jteo> aramdune: wait, nvm. it isn't in the README.
[16:33:35] <aramdune> then where is it?
[16:33:54] *** trede has joined #opensolaris
[16:35:30] <jteo> aramdune: certain putbacks exist in Solaris Express CR (latest nevada builds), and the open source OpenSolaris source code you have builds against those.
[16:35:37] <jteo> aramdune: that's my guess.
[16:36:18] <aramdune> that means it will not build against nexenta...
[16:37:09] <jteo> aramdune: you could try.
[16:38:27] *** nyati has quit IRC
[16:39:47] <kimc> if you're smart you know what s/cu/to is ;)
[16:42:37] <aramdune> me? umh... sed -e 's/cu/to' wrongfull_text_stream ? :P
[16:42:49] <kimc> :)
[16:43:37] <kimc> command garbled..
[16:45:57] *** splitscreen has left #opensolaris
[16:46:09] *** gtc_ has joined #opensolaris
[16:47:10] *** Tpenta has quit IRC
[17:07:35] *** __schily_ has joined #opensolaris
[17:09:54] *** jsubl2 has quit IRC
[17:20:10] *** jsubl2 has joined #opensolaris
[17:20:44] *** aramdune has quit IRC
[17:24:07] *** _schily__ has quit IRC
[17:34:53] *** besonen_mobile has joined #opensolaris
[17:52:27] *** LeftWing has quit IRC
[17:54:09] *** LeftWing has joined #OpenSolaris
[18:16:52] *** Teltariat has quit IRC
[18:33:14] *** vmhobbes has joined #OpenSolaris
[18:34:12] *** gtc_ has quit IRC
[18:34:21] *** gtc_ has joined #opensolaris
[18:45:58] *** jsubl2 has quit IRC
[18:50:00] *** Mdx4 has quit IRC
[19:04:50] *** deny__ has joined #opensolaris
[19:13:22] *** deny_ has quit IRC
[19:16:42] *** ThEbUtChE has quit IRC
[19:18:15] *** oooo|en_US has quit IRC
[19:18:58] *** vmhobbes has quit IRC
[19:23:25] *** oxygene has quit IRC
[19:23:29] *** oxygene has joined #opensolaris
[19:25:57] *** LordKing has joined #opensolaris
[19:31:17] *** Disorganized has joined #OpenSolaris
[19:32:53] <deny__> does anyone here know what i could put in the dhcp.conf to direct internet traffic to my web cache server?
[19:34:17] <sickness> i'm back
[19:35:43] <deny__> welcome
[19:35:52] <deny__> help me now, could u?
[19:36:49] <g4lt-mordant> you can't use dhcp.conf.  you need to use NAT
[19:36:56] <sickness> mmm you should do a man dhcp_options or something like that, lemmesee
[19:37:16] <sickness> g4lt-mordant: well, there's also a documented dhcp option, I don't know how a client could use that info, but there's the specs to do that...
[19:37:38] <sickness> (even if I agree that the best way is a forced/stealth redirect trough NAT)
[19:38:16] <g4lt-mordant> crescent hammer it is, then :/
[19:38:42] *** deather_ is now known as deather
[19:39:23] <sickness> site-local" option 252 ("auto-proxy-config")
[19:39:42] <sickness> (4th result on google :)
[19:40:23] *** Mdx4 has joined #opensolaris
[19:40:37] *** axisys has joined #opensolaris
[19:40:38] *** Mdx4 has quit IRC
[19:40:50] *** gtc has joined #opensolaris
[19:40:50] <sickness> http://en.wikipedia.org/wiki/Web_Proxy_Autodiscovery_Protocol
[19:41:37] <axisys> is sun screen equivalent to openbsd's pf? i think ipf does not offer same level granularity as pf does on opensolaris.. am i correct?
[19:41:51] <sickness> yeah
[19:42:18] <sickness> ipf is almost the same in text files configurations, but has less features
[19:42:49] <sickness> I'd still prefer openbsd+pf as a firewall over solaris+ipf
[19:43:02] <sickness> but for a lot of common tasks there should be no real difference :)
[19:46:01] <axisys> sickness: is there a pf for solaris or even possible to compile the openbsd's pf on solaris?
[19:46:35] <sickness> mmm I don't know if there's a porting effort for it, it could be surely portable since there's a port for freebsd...
[19:47:00] <axisys> also is there a benefit on sun screen?
[19:47:16] <Auralis> sunscreen can do bridging on solaris
[19:47:27] <Auralis> and its a OEM checkpoint basicaly
[19:47:48] <axisys> Auralis: sounds like it give better security than ipf ?!
[19:47:56] <quasi> sickness: I think there is some work under way, but it isn't simple because pf needs some kernel hooks iirc
[19:48:28] <axisys> now is JASS next version of sunscreen?
[19:49:16] <quasi> no
[19:49:21] *** o2addict has joined #opensolaris
[19:49:22] <sickness> quasi: heh, I'd like to see newz about that I'd be happy to try something even early alpha =)
[19:49:48] *** oxygene has quit IRC
[19:50:02] <quasi> sickness: it has been quite a while since I've seen anything - I'd suggest googling
[19:50:25] <sickness> yeah, well for carp there's the ucarp portable implementation at least :)
[19:51:04] <axisys> so can i use sunscreen instead of ipf on sol 10 u2?
[19:51:16] <trygvis> can cdrw write DVD ISOs?
[19:51:19] <axisys> or am i comparing apple w/ orange?
[19:51:46] <axisys> trygvis: it sh'd be able to .. i did it before on my ultra 20
[19:52:14] <sickness> axisys: good question, I'd like to know that too...
[19:52:47] <trygvis> coolio
[19:52:55] <Auralis> sunscreen was droped from solaris with solaris 10, it is included in solaris 9. i think you can buy sun screen for sol10, not sure
[19:53:36] <axisys> Auralis: it is replaced by something? ipf?
[19:53:39] <sickness> oh, so stock solaris10 has no integrated firewall?
[19:53:43] *** Mdx4 has joined #opensolaris
[19:53:54] <axisys> sickness: i know ipf comes w/ sol 10
[19:54:04] <quasi> sickness: ipfilter
[19:54:09] <axisys> but not sure if ipf replaced sunscreen
[19:54:11] <sickness> k
[19:55:33] <axisys> is ipf a replacement of sunscreen?
[19:55:55] <soultan> http://tinyurl.com/y7z8r3
[19:58:00] *** gtc_ has quit IRC
[19:58:54] *** Disorganized has quit IRC
[19:59:23] <Auralis> ipf replaced sunscreen in sol10 yes
[19:59:30] <axisys> another question .. is openbgpd comparable to SUNWzebrau/SUNWzebrar?
[19:59:43] *** Kernel86 has quit IRC
[19:59:54] *** Disorganized has joined #OpenSolaris
[20:00:26] <axisys> may be pf to ipf is what openbsd to SUNWzebra?!
[20:01:00] <axisys> Auralis: thnx
[20:01:29] *** n0rus has joined #Opensolaris
[20:02:15] <n0rus> How do I add users with "useradd" in a non-global zone? UX: useradd: ERROR: Unable to create the home directory: Operation not applicable. I believe that I have to edit auto_master, auto_home and vfstab. But could someone, please, guide me a little bit here?
[20:03:32] <jamesd> n0rus, its the same as in the global zone.  http://uadmin.blogspot.com/2005/02/youre-never-far-from-home.html
[20:04:18] <n0rus> ok, thank you. I just thought that I need to make a separate writeable dir in zonecfg for my zone
[20:04:37] <jamesd>  /etc is writable in zone configs
[20:04:57] <n0rus> No, I ment home dirs for the users, like /home
[20:05:00] <axisys> n0rus: your /home could be part of / of the local zone
[20:05:06] <n0rus> right
[20:05:17] <axisys> which u have full access from w/in the zone
[20:05:23] <n0rus> ok
[20:05:34] <axisys> n0rus: i just create a user.. no problem.. as root inside the zone
[20:06:27] <axisys> n0rus: check this out http://www.rafb.net/paste/results/iGFsNh30.html
[20:06:35] <quasi> n0rus: you could choose to create homedirs elsewhere - /export/home is pretty common
[20:06:48] <n0rus> yeah, that's what I have in the global zone
[20:07:19] <axisys> n0rus: if your local zone is on top of a zfs .. it does not matter where ur home dir is
[20:07:49] <n0rus> no, it's release 1/06. there's no zfs here
[20:07:53] <axisys> or follow what quasi says
[20:08:01] <n0rus> what I'm already doing....
[20:08:02] <axisys> n0rus: :-(
[20:08:17] <quasi> n0rus: -d /export/home/luser ?
[20:08:27] <n0rus> just a second:P I'm working on it here
[20:09:11] <axisys> n0rus: that's the cool w/ irc.. u get more help than u can take .. lol
[20:09:17] <n0rus> hehe
[20:09:36] <n0rus> ok, fixed now
[20:09:43] <n0rus> thanks for the help
[20:10:49] <n0rus> I was thinking about upgrading this server to 6/06 though. I've already been playing with zfs on my laptop which is running b49
[20:11:02] *** solaris-user has joined #opensolaris
[20:13:16] *** Joker^ has joined #opensolaris
[20:13:23] *** Joker^ has quit IRC
[20:13:36] <axisys> n0rus: u r braver than I am.. I am thinking about swithc to sol 10 from ubuntu on my laptop
[20:14:00] * delewis can't wait to throw Solaris on his Ferrari
[20:14:21] <axisys> delewis: like to hear your experience..
[20:14:44] <delewis> axisys: a few people have already documented Solaris on the 4000 series
[20:14:48] <n0rus> axisys: the last time I tried to install OpenSolaris (and it was build 30 something), I couldn't even get to the installer. But now the hardware support is wider and so the installation went almost painfree
[20:14:55] <delewis> but no one seems to have installed on a 5000 yet
[20:14:59] <n0rus> And I have an ASUS W2V
[20:15:26] <axisys> on my laptop hp nc6220 (not listed in Laptop HCL) all button (media) except wifi button works right ways w/ ubuntu
[20:15:36] *** o2addict is now known as oxygene
[20:15:37] *** IRCMonkey has joined #opensolaris
[20:16:03] <delewis> well, going the Ferrari route was a lot cheaper than dealing with Tadpole or Naturetech
[20:16:04] <n0rus> I still can't get my internal speakers to work though
[20:16:16] *** IRCMonkey has quit IRC
[20:16:16] <axisys> wifi works w/ ubuntu just the button does not lit up when active
[20:16:21] <delewis> Naturetech came back with a $5000 USD quote on the base model of their cheapest laptop.
[20:16:32] <delewis> (which was UltraSPARC-IIe with 256MB of memory)
[20:16:41] <axisys> delewis: ouch!
[20:16:49] <quasi> sparc ain't cheap
[20:16:56] <solaris-user> ouch!
[20:16:56] <n0rus> no shit:P
[20:16:59] <delewis> of course, they had a deal with Sun (which has been cancelled now) to sell a midrange model for $3.4k
[20:17:02] <delewis> rip-off.
[20:17:18] <axisys> quasi: guess not
[20:17:23] <delewis> SPARC portable solution vendors only care about selling to corporations or the military.
[20:17:42] <n0rus> yeah, it's mostly scientists and military
[20:17:45] <delewis> and from what I hear, Tadpole almost exclusively only sells to military nowadays.
[20:17:52] <delewis> which is sad.
[20:18:07] <n0rus> like you would have payed 5k:)
[20:18:21] <delewis> if they would just drop the prices by a few thousand and actually care about the hobbyist market, they'd have a huge increase in sales.
[20:18:32] <delewis> n0rus: I would've paid $3.5k at the most.
[20:18:39] <axisys> delewis: i second u
[20:18:50] <n0rus> well, 3.5 sure. but not 5
[20:19:17] <axisys> w/ opensparc it may come down in price sometime soon i think
[20:20:20] <delewis> and what's interesting is that Naturetech/Tadpole/etc. all seem to have the same basic design for all their laptops.
[20:20:39] <delewis> same chassis, same board, etc.
[20:21:12] <axisys> best think would be to win a coolthread.. hehe
[20:21:34] <delewis> until the T2 comes out, the T1 wouldn't be suitable for a portable solution.
[20:22:03] <n0rus> axisys: http://lj.deformica.com/sol_desktop.jpg
[20:22:09] <delewis> terrible floating point performance, and the question of whether or not a multi-threaded workload would be present on a laptop is still not determined.
[20:22:47] <delewis> and Sun doesn't seem to be interested in portable solutions, anyway -- they cancelled the Sun Ultra 3
[20:23:46] <axisys> n0rus: pretty cool
[20:23:53] <n0rus> :)
[20:24:21] <solaris-user> one day i will have one sparc-note, but not at this prices --_-
[20:24:42] <delewis> solaris-user: it hasn't changed in 10 years, so I doubt it'll be changing anytime soon.
[20:24:54] <delewis> portable SPARC vendors will convert to AMD64 lines before they drop SPARC prices
[20:24:59] <delewis> as Tatung did
[20:26:05] *** Mdx4 has quit IRC
[20:26:14] <axisys> n0rus: u have a tv tuner on it?
[20:26:22] <n0rus> yep
[20:27:02] <axisys> n0rus: and it works w/ opensolaris i take?
[20:27:17] <delewis> doubtful.
[20:27:24] <delewis> I don't think there's a bktr driver
[20:27:37] <solaris-user> n0rus: which model?
[20:27:47] <n0rus> Haven't tried yet actually. The tuner wasn't even listed in scanpci
[20:27:52] <n0rus> which model of what?
[20:27:56] <n0rus> tuner?
[20:28:01] <solaris-user> yeap
[20:28:08] <delewis> there was a tuner application years ago for Solaris written by Sun
[20:28:14] <delewis> but I doubt that was ever ported to Solaris/x86
[20:28:27] <n0rus> I can't recall now
[20:28:40] <delewis> and it probably only supported a subset of tuners -- i.e. the ones Sun shipped with workstations
[20:29:02] <n0rus> I'm not using this tuner anyway. Even when I'm in Windows
[20:29:09] <delewis> hehe
[20:29:10] <n0rus> have a dual boot
[20:29:21] <delewis> I plan on triple booting my Ferrari -- Solaris, Windows, and Slackware
[20:29:31] <n0rus> hardcore
[20:29:32] <axisys> what is `scanpci' equivalent on sol 10 x86 u2?
[20:30:10] <axisys> got it
[20:30:11] <sickness> n0rus: where did you find your desktop background image?
[20:30:18] <delewis> axisys: /usr/X11R6/bin/scanpci
[20:30:20] <n0rus> sickness: it was bundled
[20:30:22] <axisys> it is scanpci.. wasn't in my path :-)
[20:30:32] <sickness> oh, so I should have it too in a b49 ?
[20:30:48] <n0rus> :P
[20:30:49] <delewis> sickness: definitely.
[20:30:58] <sickness> I'll try to search for it :)))
[20:31:00] <delewis> I forget when the new backgrounds were integrated, but they are certainly click.
[20:31:03] <delewis> slick, rather.
[20:31:14] <n0rus> sickness: I could upload it for you
[20:31:15] <delewis> the backgrounds in JDS previously, well.. sucked.
[20:31:28] <delewis> standard Gnome stuff, and a few Java backgrounds.
[20:31:33] *** Mdx4 has joined #opensolaris
[20:31:42] * delewis wishes the CDE Sun logo background was still around
[20:31:48] <sickness> nope, I'll search for it tnx :)
[20:31:49] <delewis> but I think that got lost in Solaris 8 or 9
[20:31:57] <n0rus> sickness: ok, as you want
[20:32:41] <axisys> sickness: u r running sol 10?
[20:33:43] <sickness> snv49
[20:34:56] <axisys> n0rus: have u tried compiz? i am using it on my ubuntu.. wonder if opensolaris has it?!
[20:35:03] <n0rus> what's that?
[20:35:11] <axisys> 3D acceleration
[20:35:37] <axisys> http://en.wikipedia.org/wiki/Compiz
[20:36:58] <axisys> i can never see myself move away from 3D accel.. i love it.. have been using it for couple months
[20:37:50] <n0rus> I would rather be switching between windows like Tom Cruise in Minority Report.
[20:38:35] <sickness> LOL
[20:38:52] <n0rus> now that was cool
[20:38:56] <axisys> n0rus: check this out http://www.youtube.com/watch?v=-CgqWlX_GsI
[20:38:59] <n0rus> it looked so at least
[20:39:00] *** loke has quit IRC
[20:39:06] <axisys> n0rus: i can do all those and more
[20:41:31] *** Fish- has joined #opensolaris
[20:44:02] *** lisppaste3 has joined #opensolaris
[20:44:44] <trygvis> grr .. the solaris 10 installer just hangs after giving the initial copyright blurb
[20:44:53] <trygvis> how can I get some more debugging info?
[20:45:54] *** coffman has joined #opensolaris
[20:47:28] <axisys> boot -m verbose i think
[20:48:15] <trygvis> no can do, it's a blank machine.. but I gave it -s on the command line as that's what the failsafe grub entry has
[20:48:21] <trygvis> but didn't give any clues
[20:49:38] *** Fish has quit IRC
[20:49:54] <axisys> can u add -a ?
[20:50:22] <axisys> it will be interactive at that point.. and u could just hit enter to take defaut.
[20:50:25] <trygvis> on the kernel line? sure
[20:50:38] <trygvis> interactive?
[20:50:43] <axisys> except the /etc/system.. for that u could choose /dev/null
[20:51:22] <axisys> if it boots at that time then all you do is inspect the /etc/system for any error.. or just comment all using `*'
[20:51:28] <n0rus> finished watching the video. yeah, there were some interesting things
[20:51:37] <axisys> n0rus: heh
[20:52:24] <solaris-user> question: i need one app to record my screen ( like record all i doing )  anyone knows one apps for solaris?
[20:53:33] <axisys> solaris-user: there is a screen capture command.. cant remember
[20:53:53] <trygvis> axisys: nope, still won't start
[20:54:23] <axisys> trygvis: did it take -a
[20:54:47] <trygvis> what do you mean? I gave it -a
[20:55:31] <n0rus> trygvis: is your installer hangs, why not use the console mode install?
[20:56:04] <n0rus> if, I ment
[20:56:33] <trygvis> dudes, the installer won't even start
[20:56:42] <trygvis> it just hangs after the copyright message
[20:56:56] <n0rus> what machine is it?
[20:57:17] <trygvis> some intel machine .. I've installed solaris on it before
[20:57:18] *** Mdx4 has quit IRC
[20:57:35] <n0rus> did you check RAM?
[20:58:32] <trygvis> not explicitly but it has been running linux just fine for a while
[20:59:44] <n0rus> well, I've had a simlar problem. I had a machine running linux. Installed solaris on it (everything went fine). Then rebooted and never got a video signal. It turned out one of my RAMs went wacko
[20:59:49] <n0rus> similar
[20:59:54] <axisys> trygvis: u have a monitor connected to it? i am thinking if need to redirect the output
[21:00:08] <trygvis> yes I have a monitor connected
[21:00:27] <trygvis> and I can't find my null modem cable
[21:01:27] <axisys> trygvis: can u get to your bios and check if console is derected to screen and not ttya
[21:02:06] <trygvis> that's not likely possible on this machine .. it's not a server
[21:02:15] <trygvis> it's a cheap office machine
[21:03:00] <trygvis> it seems to hang after: cpuid 0: initialized cpumod: cpu.generic
[21:03:11] <trygvis> (now I gave it /dev/null as /etc/system)
[21:03:26] <axisys> trygvis: ok..
[21:03:41] <axisys> trygvis: that didnt help i take?!
[21:04:13] <trygvis> no, it just hung there
[21:04:20] <axisys> hmm
[21:05:06] <axisys> u may try with another live cd if u have one around..
[21:05:37] <axisys> to make sure the problem is consistent.. otherwise could be something in ram like n0rus suggested
[21:05:55] <trygvis> it's not the ram, I'm sure of that
[21:06:00] <axisys> ok
[21:06:32] <trygvis> I remember reading something about someone needing a new bios to load the new microcode instruction set into the cpu before beeing able to boot
[21:07:14] <n0rus> you sure you weren't dreaming?:)
[21:08:15] <axisys> solaris-user: have u tries Alt+Printscreen?
[21:08:24] <axisys> s/tries/tried/
[21:08:59] <n0rus> axisys: I think he's talikng about video capturing a la Camtasia Studio for Windows
[21:09:21] <axisys> n0rus: i c
[21:10:45] <solaris-user> yeo like Camtasia
[21:10:47] <solaris-user> :)
[21:10:50] <solaris-user> I need this
[21:10:55] *** LordKing has quit IRC
[21:11:00] <axisys> solaris-user: this java tool seems to does that unless i misread it http://java.sun.com/products/java-media/jmf/2.1.1/solutions/JVidCap.html
[21:12:32] <trygvis> hehe .. there are microcode updates for the bios ..
[21:12:46] <trygvis> now the problem is that I don't have a single floppy in the house
[21:14:47] <trygvis> crap, I'm giving up
[21:15:36] <n0rus> ikke gi opp:P
[21:16:12] <trygvis> o.O
[21:16:31] <trygvis> 1) I don't have a floppy, 2) neither do I have a windows to extract the files
[21:16:59] <n0rus> 1) it sucks 2) 1)
[21:17:07] <icon> trygvis: have a work box with a cd burner?
[21:17:14] <icon> just make a bootable cd
[21:18:25] *** coffman has quit IRC
[21:22:12] *** bitvector2 has joined #opensolaris
[21:22:33] <bitvector2> anyone here compile and run Subversion 1.4.0 with Studio?
[21:23:39] *** kimc has quit IRC
[21:24:32] <quasi> bitvector2: uhm, yeah - I think I ended up getting it to work
[21:25:42] <quasi> bitvector2: a bit of a fight though
[21:28:17] <bitvector2> quasi: which apache did you build agaist?
[21:28:56] <quasi> bitvector2: ah, not the server bits - I only needed the client
[21:29:05] <bitvector2> lol
[21:29:08] *** gtc has quit IRC
[21:29:24] <bitvector2> you didn't suffer a fight at all then ;)
[21:29:53] <bitvector2> I'm getting weird runtime errors
[21:30:08] <quasi> precompiled httpds on solaris are cr@p anyway - I always roll my own anyway
[21:30:29] <bitvector2> like when I svn commit on one machine and then svn update on another machine- there isn't anything to be updated
[21:30:55] <bitvector2> quasi: yeah I agree, but I'm stuck having to try anything I can think of at this point
[21:31:17] <bitvector2> First I tried my own custom whizbang apache 2.2.3
[21:31:26] <quasi> bitvector2: good luck, you're going to need it ;)
[21:31:38] <bitvector2> that was full of runtime problems, now I'm trying the builtin Solaris apache2
[21:32:05] <bitvector2> btw apache 2.2.3 won't even compile with Studio
[21:32:30] <bitvector2> has the whole world just forgot that Linux isn't everything?
[21:32:55] <bitvector2> I would expect more from the Apache project
[21:34:13] <quasi> what? httpd not compiling with studio? very odd - it did last time I tried
[21:34:34] <bitvector2> sure the 2.0.x series
[21:34:52] <bitvector2> but the lastest 2.2.x - nope
[21:34:53] *** neoxed has quit IRC
[21:35:15] <quasi> I probably did trunk last
[21:35:49] <sickness> drunk :P
[21:36:32] <quasi> bitvector2: the only bits I ran into was openssl giving a bit of problems, but that was fixable
[21:38:09] <bitvector2> do you still have that trunk code laying around?
[21:38:52] <quasi> bitvector2: let me take 2.2.3 for a spin
[21:39:06] <bitvector2> can I send you or paste my build script?
[21:39:46] <quasi> bitvector2: just a recent checkout - but be warned that the auth stuff in trunk is somewhat broken
[21:39:57] *** neoxed has joined #OpenSolaris
[21:41:36] <bitvector2> #!/usr/bin/env bash
[21:41:37] <bitvector2> export PKG='httpd-2.2.3'
[21:41:37] <bitvector2> rm -rf $PKG
[21:41:37] <bitvector2> gtar -xjf $PKG.tar.bz2 &&
[21:41:37] <bitvector2> chown -R ${USER}:${USER} ${PKG}* &&
[21:41:37] <bitvector2> cd $PKG &&
[21:41:39] <bitvector2> export CC='cc' &&
[21:41:41] <bitvector2> export CFLAGS='-mt -fast -xtarget=opteron -xarch=sse2 -fround=nearest -xvector=simd' &&
[21:41:43] <bitvector2> export CPPFLAGS='-I/usr/local/include -I/usr/sfw/include' &&
[21:41:45] <bitvector2> export CXX='CC' &&
[21:41:48] <bitvector2> export CXXFLAGS=$CFLAGS &&
[21:41:49] <bitvector2> export LDFLAGS='-R/usr/local/apache2/lib:/usr/local/lib:/usr/sfw/lib' &&
[21:41:51] <quasi> stop please
[21:41:51] <bitvector2> export LD_LIBRARY_PATH='/usr/local/lib:/usr/sfw/lib' &&
[21:41:53] <bitvector2> ./configure --prefix=/usr/local/apache2 --exec-prefix=/usr/local/apache2 --enable-modules=all --enable-mods-shared=all --disable-authn-dbd --enable-authnz-ldap --disable-dbd --enable-ldap --enable-proxy --enable-ssl --with-included-apr --with-pcre=/usr/local --with-z=/usr/local --with-ssl=/usr/sfw --with-mpm=worker --with-ldap &&
[21:41:57] <bitvector2> gmake -j4 &&
[21:41:59] <bitvector2> gmake install &&
[21:42:01] <bitvector2> cd ..
[21:42:03] <bitvector2> its done
[21:42:06] <bitvector2> thats it
[21:42:51] <bitvector2> put this script in the same directory as the tarball  and run it
[21:43:16] <bitvector2> well comment out the gmake install
[21:43:50] <quasi> try it without your cflags first
[21:44:40] <icon> be very wary of enabling sse instructions on stuff that doesnt explicitly support it
[21:44:47] <icon> i cant tell you how many times thats bitten me
[21:45:11] <bitvector2> wow
[21:45:48] <delewis> why are you building Apache with SSE support?
[21:46:00] <bitvector2> I build everything with the same CFLAGS
[21:46:06] <delewis> that's silly.
[21:46:21] <delewis> the only things that benefit from SSE instructions tend to be multimedia applications
[21:46:24] <bitvector2> I have 35 projects that I've been running and not a single prob till now
[21:46:29] <icon> sse tends to break tons of stuff
[21:46:41] <delewis> bitvector2: that's a Gentoo-ish philosophy.
[21:46:43] <delewis> it's dumb.
[21:46:45] <icon> in fact, most kernel builds have it shut off explicitly because of those issues
[21:46:45] <bitvector2> so you're telling me the compiler isn't smart enough to realize that?
[21:46:58] <quasi> bitvector2: builds cleanly for a similar config without the cflags
[21:47:09] <icon> quasi: imagine that :D
[21:47:22] <bitvector2> try the same flags
[21:47:30] <andersmo> bitvector2: the compiler i smart enough to realize that and not use sse by default, but you tell the compiler to disregard that? =)
[21:47:39] <andersmo> s/i smart/is smart/
[21:48:05] <icon> bitvector2: to be perfectly honest, you dont need most of those flags.
[21:48:08] <quasi> bitvector2: no, why would I want to break it
[21:48:16] <delewis> there's just no point in enabling an instruction set (or rather trying to enable an instruction set) that'll have no effect on application performance.
[21:48:21] <icon> sunstudio is *not* gcc
[21:48:23] <bitvector2> quasi: because your test is meaning less to me
[21:48:36] <andersmo> cc -funroll-loops -fomit-instructions --omg-optimized. =)
[21:48:39] <bitvector2> it doesn't prove that my CFLAGS are bad
[21:48:41] <delewis> bitvector2: why do you want the compiler to enable SSE when building Apache?
[21:48:44] <delewis> that's stupid.
[21:48:49] <bitvector2> is just proves that your CFLAGS are good
[21:48:56] <icon> andersmo: d00d, you forgot -O65535 !
[21:49:01] <quasi> bitvector2: don't speak badly about httpd when you're doing silly things like that
[21:49:15] <bitvector2> delewis: whats stupid is your not seeing the point
[21:49:23] <bitvector2> the point is will it compile
[21:49:32] <bitvector2> not whether it is desirable
[21:49:35] <delewis> bitvector2: the point is you don't understand what SSE does.
[21:49:36] <quasi> bitvector2: httpd works just fine except for your specific choice of cflags that defy all logic
[21:49:40] <icon> bitvector2: are you an apache developer?
[21:49:50] <bitvector2> look I'm a CpE and I now damn  well was SSE
[21:49:51] <bitvector2>  it
[21:49:58] <delewis> if you understood what SSE does, you wouldn't be enabling it to build Apache.
[21:50:01] <icon> bitvector2: odds are, you havent got a clue whether any of those 'optimizations' will have any real effect
[21:50:08] <delewis> bitvector2: sorry, you aren't very convincing.
[21:50:26] <icon> it is completely possible there are bits in the apache source that could conflict with some compilier optimizations for sse support
[21:50:31] <bitvector2> nevermind folks you just want to argue
[21:51:15] <andersmo> bitvector2: they are trying to give you sound advice and share a bit of their software development experience. =)
[21:51:21] <bitvector2> I understand that
[21:51:30] <bitvector2> thank you and it is taken
[21:51:37] <bitvector2> but I just want to know if it will compile
[21:52:01] <icon> bitvector2: pop onto #apache and ask if those CFLAGS will have any odd effects
[21:52:12] <delewis> if it doesn't compile when you use certain optimizations it is not the application's fault.
[21:52:14] <icon> just because it compiles doesnt mean it will work as expected
[21:52:14] <bitvector2> I don't want to ask for someone to speculate
[21:52:17] <delewis> it's either the compiler's or the user's.
[21:52:32] <delewis> in most cases, the user's, because most compilers make it clear when not to use silly instructions.
[21:52:33] <andersmo> Usually, whether something compiles or not isn't as interesting as the question of whether the resulting binary executes correctly. =)
[21:52:34] <icon> bitvector2: exactly. talk to a developer who knows the source
[21:53:07] <andersmo> Anyway, I have no idea how much floating point math httpd has. Probably not a whole lot. =)
[21:53:19] <delewis> (for example, read the 'cc' man page how instances were -fast is a *bad* idea)
[21:53:20] <Disorganized> http://www.youtube.com/watch?v=JMvMzQ4Vu-8
[21:53:22] <icon> as a general rule, compilers generally dont need help deciding how to output reasonable bins
[21:53:26] <delewis> (-fast is a bad idea in 99% of cases)
[21:53:33] <bitvector2> delewis I already read it
[21:53:44] <icon> bitvector2: then why are you using it?
[21:53:45] <bitvector2> delewis I've already spoken with Kuldip
[21:53:55] <bitvector2> delewis don't assume that I haven't done my research
[21:54:10] <quasi> delewis: not necessarily agreed on -fast being bad - http://developers.sun.com/sunstudio/articles/amdopt.html
[21:54:39] <delewis> quasi: the majority of applications when built with -fast perform worse than if you don't use optimizations, period.
[21:54:51] <delewis> though, there are some rare cases where -fast is useful.
[21:54:58] <delewis> emphasis on "rare"
[21:55:03] <quasi> delewis: I somehow doubt that
[21:55:18] <delewis> quasi: try buiding SDL or lame with -fast
[21:55:53] <quasi> delewis: especially given that -fast is most of all a hack from sun to get over a limit in spec of how many cmd line params you're allowed to use
[21:56:18] <quasi> delewis: let's talk servers instead
[21:56:25] <delewis> or just a macro for slackers to use to keep from reading the man page about various optimization options :-)
[21:56:36] <andersmo> Optimizing for size instead of performance is often a good idea for performance on many apps nowadays, afaik? I-caches are good for you. =)
[21:57:04] <delewis> quasi: servers have no use for the majority of -fast options
[21:57:18] <bitvector2> delewis why do you always generize?
[21:57:34] <icon> bitvector2: like your CFLAGS?
[21:57:38] <bitvector2> how about a test and see attitude
[21:57:43] <icon> (sorry i couldnt resist) ;)
[21:57:56] <delewis> bitvector2: I have tested, as I said I specifically cited two applications -- lame and SDL
[21:58:08] <bitvector2> and so you generlize from 2 to many?
[21:58:10] <bitvector2> nice
[21:58:11] <delewis> when compiled with -fast they perform horribly compared to if they weren't even built with optimizations
[21:58:14] <icon> lame is an interesting example, considering the amount of pure processing it does
[21:58:21] <delewis> and those are multimedia applications that *could* benefit from the optimizations
[21:58:34] <bitvector2> so don't compile those apps that way and don't blaspheme -fast for everything
[21:58:39] <delewis> a server application couldn't even benefit from them, so what makes you think they'd perform any better?
[21:58:56] <bitvector2> how the hell do you know what all server apps are doing?
[21:59:15] <icon> id honestly hate to burst your bubble bit, but if you are needing to optimize apache to squeeze more performance out of it, you need new hardware
[21:59:26] <quasi> delewis: I'll have to go do some tests first - but -xtarget=native and -xO5 look like good candidtes
[21:59:29] <icon> httpd should *not* be under heavy load
[21:59:38] <icon> webservers are typically disk and mem bound, not by cpu
[21:59:43] <quasi> icon: that is BS
[21:59:45] <delewis> quasi: yes, those alone seem to be OK
[22:00:06] <delewis> but the rest of the options that -fast enables impact performance
[22:00:19] <bitvector2> benchmark results?
[22:00:37] <icon> quasi: not at all. i work with very large farms every day. with respect to httpd you will almost always see disk io and mem bottlenecks before cpu
[22:00:45] <delewis> on the order of about 3 magnitudes slower
[22:00:50] <delewis> in the case of lame
[22:00:59] <delewis> SDL was similar, IIRC.
[22:01:00] <icon> thats assuming typical serving, and not more cpu intensive bits like heavy rewrites and compression
[22:01:33] <delewis> but also this was on SPARC
[22:01:39] <delewis> results on x86 might differ
[22:01:40] <quasi> icon: right, that part I don't disagree on - it was the "httpd should *not* be under heavy load"
[22:01:51] <delewis> UltraSPARC-III at that, so -xprefetch is enabled
[22:01:54] <icon> quasi: ahh sorry, i should have qualified that :)
[22:02:03] <delewis> though, I doubt that would have any effect
[22:02:04] <icon> you shouldnt see more cpu load that then others
[22:02:10] <icon> s/that/than/
[22:03:00] <icon> besides, theres not a damned thing that apache would benefit from sse other than unintended consequences
[22:03:01] <quasi> icon: then we agree (although iowait may look like high server load)
[22:03:10] <icon> and fround makes me squirm
[22:03:13] <icon> yup
[22:03:28] <delewis> icon: that was my point, originally :-)
[22:04:42] <icon> bitvector2: id look at iostat and vmstat under load before i would start considering compile time optimizations - odds are you are just making more work for yourself, and taking a big risk on unspecified behavior
[22:04:53] <andersmo> bitvector2: generally, at low levels, lots of server software do similar things. They mostly do integer processing, for one thing, and do lots of decision-making and branching. Also, they are often memory-intensive and IO-intensive. Lots of CPU time is actually spent waiting for memory, and lots of the server software's time is spent in the operating system kernel, performing networking, disk i/o and IPC. This is what guided the design of the
[22:04:55] <icon> bitvector2: you know, what typical admins will do
[22:05:18] <icon> precisely
[22:05:52] <quasi> andersmo: you got cut off at "This is what guided the design of the"
[22:06:09] <andersmo> ah, sorry 'bout that. ..."design of the niagara processor, for instance. <- Now, _that_ is a generalization."
[22:06:20] <bitvector2> icon, delewis, thanks but I just ansered my question
[22:06:40] <bitvector2> I was accidentally using a stale libtool.  my CFLAGS work fine
[22:07:17] <bitvector2> icon do you know what the iowait field means?
[22:07:21] <andersmo> bitvector2: Problems with CFLAGS usually don't manifest until after compilation, but then the bugs can be pretty obscure. =)
[22:07:36] <bitvector2> does it mean that the CPU is waiting on IO?
[22:07:51] <delewis> and then you spend lots of time disassembling or just re-building completely (without ever knowing at caused the bug)
[22:08:05] <delewis> what*
[22:08:13] <bitvector2> andersmo: thanks thats why running ab
[22:08:54] *** deny__ has quit IRC
[22:09:27] <andersmo> random SIGSEGVs, wrong (but almost right, maybe close enough to be unnoticeable at first, but that depends on the app) results in floating-point math because SSE isn't exactly the same thing as x87 math, etc...
[22:09:31] <Error_404> hmm... this is cute
[22:09:51] <Error_404> SUNWappserver isn't asking for authentication for webdav
[22:10:38] <icon> bitvector2: its the percentage of time spent waiting for io, yes
[22:10:52] <icon> bitvector2: assuming you are looking at %'s in top
[22:11:14] <icon> thats what blocking io does...
[22:11:34] <quasi> prstat -m gives a much more correct picture, but takes a bit of practice to read
[22:11:42] <icon> *nods*
[22:11:45] <axisys> i have a libpcap.a and libpcap.so on same dir.. even if i use --with-pcap-lib=/opt/csw/lib as a valid option with configure.. it still says -lpcap not found
[22:12:07] <icon> http://developers.sun.com/solaris/articles/prstat.html
[22:12:14] <icon> (in case anyone was curious)
[22:12:59] <icon> axisys: have you checked your LD_LIBRARY_PATH ?
[22:13:14] <axisys> icon: i rather not touch that env
[22:13:23] <quasi> icon: "Explaining the meaning and implications of all these columns is beyond the scope of this paper" ;)
[22:13:30] <icon> haha
[22:13:34] <icon> thats cute, i didnt see that bit
[22:14:16] <quasi> solaris internals and especially the performance tuning part goes into excruciating detail
[22:14:19] <icon> axisys: have you worked with solaris a bit?
[22:14:22] <icon> it does
[22:14:30] <bitvector2> icon can you tell me why Solaris 10 shipped with the iowait field blank?
[22:14:35] <icon> my internals book was on 8, i need to get a new one, but thats an excellent book
[22:14:39] <axisys> icon: just 5 yrs.. not much
[22:15:14] <quasi> icon: the new one is very nice
[22:15:15] <icon> axisys: no worries, just wanted to be sure i didnt have to bring out the ldconfig (or lack thereof) stick
[22:15:34] <axisys> icon: u mean crle?
[22:15:37] <icon> axisys: give it a shot and see if that fixes it, ive had some trouble like that in the past
[22:15:43] *** drio has joined #opensolaris
[22:16:18] <icon> axisys: depends on whether you are using gcc or sunstudio
[22:16:33] <axisys> icon: i guess i could.. but wondering if there is a gcc option to point to the library directly instead of just path
[22:16:36] <icon> iirc gcc ignores crle
[22:16:54] * quasi is not the biggest fan of using crle as it impacts system wide
[22:16:56] <delewis> gcc has nothing to do with crle
[22:17:08] <andersmo> axisys: have you stared hard at the configure.log? It lists the complete compiler command lines, and could reveal brokenness or unexpected assumptions in the configure script. =)
[22:17:13] <icon> okay, well gnu ld then
[22:17:15] <delewis> crle is for run-time linking, whereas gcc is concerned with *compile-time* linking.
[22:17:34] <icon> apologies, i was over simplifying
[22:17:47] <axisys> andersmo: let me look at the log
[22:18:16] <delewis> axisys: why do you want to point to just the library, specifically, and not the library's path?
[22:18:22] <delewis> it makes no difference to the linker
[22:18:25] <axisys> delewis: crle also allows to add a library path and then compile with gcc should just find it.. not?
[22:18:32] <delewis> axisys: no
[22:18:36] <delewis> crle is a *run-time* linker
[22:18:37] <icon> axisys: depends on your linker
[22:18:42] <delewis> or rather
[22:18:46] <delewis> modifies the run-time linker configuration
[22:18:50] <andersmo> I've found a bit of configure.log reading very enlightening many times when I've been struggling with compiling software and trying to coax the compiler and linking into using the right headers and libraries.
[22:18:51] <axisys> delewis: becasue i have both static and shared in same path
[22:19:09] <delewis> axisys: gcc should use the shared ones, unless you specify static linking
[22:19:16] <icon> axisys: mod your ld lib path, or link statically
[22:19:23] <delewis> ugh
[22:19:24] <delewis> no
[22:19:29] <axisys> icon: nah
[22:19:50] <delewis> specify the library paths with -L and specify that you want to embed the run-time linker paths in the binaries via -Wl,-R<library paths colon separated>
[22:19:55] <delewis> you should never, never have to toy with crle
[22:19:56] <icon> delewis: last i checked ldconfig isnt part of /usr/sfw... how will gnu ld find the specific path?
[22:20:16] <icon> ahh, mod the Makefile then... cheater :)
[22:20:27] <axisys> LDFLAGS="-L/opt/csw/lib -R/opt/csw/lib" ./configure --with-pcap-headers=/opt/csw/include --with-pcap-lib=/opt/csw/lib --with-gnu-ld
[22:20:37] <delewis> icon: ldconfig should have nothing to do with compile-time
[22:20:43] <axisys> delewis: that is how i ran it.. but it did not find the dynamic lib
[22:20:49] *** besonen_mobile has quit IRC
[22:20:56] <delewis> like crle, it just modifies the run-time linker configuration
[22:21:01] <delewis> or it should, anyway.
[22:21:04] <andersmo> 'LD_OPTIONS="-R/opt/ITEAdb/lib:/opt/ITEAcyrus/lib" ./configure CC=cc CFLAGS="-O" \ ...' <- hacks like that might be necessary, for instance. =)
[22:21:10] <icon> delewis: hrmm. i was under the impression ldconfig was used for compiletime as well
[22:21:19] <delewis> icon: it shouldn't be
[22:21:24] <delewis> if you want gcc to search various library paths by default
[22:21:29] <delewis> you need to edit the gcc configuration file
[22:21:29] <axisys> andersmo: what does -O do ?
[22:21:31] <andersmo> ah, I see axisys already does that. =)
[22:21:36] <icon> optimizations
[22:21:41] <andersmo> axisys: generic optimization.
[22:21:41] <delewis> (I forget what it is, but there's a place to search various paths by default)
[22:21:44] <axisys> icon: that is what i thought
[22:21:47] <icon> generally you are save with -O2 at the most, any higher and youll see issues
[22:21:51] <icon> s/save/safe/
[22:21:53] <bitvector2> damn those CFLAGS sure are broke - 1700 requests per second, and 38 Mb/sec
[22:22:00] <axisys> icon: yeap
[22:22:31] <axisys> delewis: what u thing about these options.. any mod u suggest?
[22:22:50] <axisys> LDFLAGS="-L/opt/csw/lib -R/opt/csw/lib" ./configure --with-pcap-headers=/opt/csw/include --with-pcap-lib=/opt/csw/lib --with-gnu-ld
[22:23:30] <icon> dont forget to use -I/opt/cw/include\
[22:23:43] <icon> (drop the \, i slipped)
[22:24:14] <axisys> icon: --with-pcap-headers took care of that
[22:24:31] <icon> axisys: until you add another option :)
[22:24:50] *** rawtaz_ has left #opensolaris
[22:24:58] <delewis> I never --with-* options to specify includes
[22:25:01] <axisys> icon: hehe
[22:25:06] <delewis> I generally use CPPFLAGS unless the configure script ignores CPPFLAGS
[22:25:22] <delewis> any particular you're wanting to use --with-gnu-ld over the Sun-supplied ld?
[22:25:23] <axisys> delewis: ./configure --help suggested that.. hence the use
[22:25:34] <axisys> delewis: not really
[22:25:40] <delewis> axisys: don't then :-)
[22:25:43] <gdamore> wow.  i just discovered something scary.  debian linux on my cobalt raq2 is a  faster NFS server than Solaris 10 (FCS) loaded on my 500MHz SunBlade 100.
[22:25:44] <icon> CXXFLAGS?
[22:25:48] <axisys> delewis: just wanted to get it find the damn lib
[22:25:49] <gdamore> the SunBlade is basically idle.
[22:25:58] <delewis> the GNU linker doesn't work on most platforms
[22:26:06] <axisys> delewis: i tried w/o that first
[22:26:26] <delewis> gdamore: scary.
[22:26:33] <axisys> delewis: no luck hence tried w/ it.. i will remove that.. which is back to where i am .. any other sugg?
[22:26:49] <gdamore> the cobalt is a 250MHz machine, with only 128MB ram.  the SB100 is a 500MHz with ~1GB ram.
[22:26:55] <delewis> though, IIRC, the Linux NFS implementation takes a few shortcuts for better performance at the sake of data corruption, etc.
[22:27:11] <delewis> there was a paper written awhile back that documented the "shortcuts"
[22:27:21] <icon> meh, linux nfs, that brings back bad memories
[22:27:24] <gdamore> i wonder how well netbsd on the beast will perform.  but it made up my mind that for moderate NFS I'll use the cobalt
[22:27:29] <delewis> not to mention the Linux NFS implementation also performs better with the Linux NFS client
[22:27:38] <Auralis> and don't forget that the sb100 is a very crap hardware
[22:27:40] <delewis> (both are aware of the shortcuts) :-)
[22:27:41] <gdamore> I'm using a _Solaris_ client.
[22:27:48] <gdamore> (Ultra 20 running S10 3/05)
[22:27:52] <delewis> gdamore: still, the shortcuts still apply, I'm sure.
[22:28:05] <gdamore> the shortcuts are probably lazy sync.
[22:28:34] <gdamore> i would have thought the sb100 @ 500MHz would handily outperform the older MIPS machine.
[22:28:45] <icon> gdamore: have you tried r/w tests, and not just ro?
[22:28:49] <icon> they could be doing delayed writes
[22:28:54] <gdamore> this was a write test, actually.
[22:29:07] <gdamore> hmm.... i should try a read test.  hang on.
[22:30:00] *** Tpenta has joined #opensolaris
[22:30:24] *** ChanServ sets mode: +o Tpenta
[22:31:28] *** n0rus has quit IRC
[22:31:49] *** xzilla has quit IRC
[22:31:52] *** xzilla has joined #opensolaris
[22:32:07] <gdamore> hmm... the sb100 handily beat the cobalt on read performance.  i think the write test was the result of delayed writes.
[22:32:22] <icon> *nod*
[22:32:35] <icon> that could get fun during a large transfer and a nice little kernel panic :D
[22:32:51] <gdamore> yeah.
[22:33:13] <icon> so how big was the difference on writes?
[22:33:16] *** xzilla has quit IRC
[22:33:25] <gdamore> i want to setup a home file server for very moderate use, and i have some cobalt hardware lying around.  the only thing is that i occasionally build NetBSD trees, so I want something that isn't going to be too slow for that.
[22:33:28] <icon> what were the disk types on there too?
[22:33:52] <icon> ide or scsi in the blade?
[22:33:57] <gdamore> the cobalt has a 40GB 5400RPM IDE drive.  the write performance was roughly 2x.
[22:33:59] <gdamore> ide.
[22:34:09] <icon> ouch
[22:34:14] <gdamore> the blade has a stock bottom end ide drive, whatever it orginally shipped with.
[22:34:22] <icon> couldnt tell you
[22:34:23] <gdamore> but on read, the cobalt was 3x slower.  so go figure
[22:34:31] <icon> well
[22:34:34] <icon> ide is pretty cpu intensive
[22:34:51] <gdamore> it can be.  depends on whether the controller  and driver support DMA or not.
[22:35:14] <icon> even with dma theres still a bit of overhead compared to sata or scsi
[22:35:21] <gdamore> actually, i have to stand corrected, i have a 2nd seagate ide drive in the SB100.
[22:35:37] <gdamore> really?  i've never looked at IDE very closely.
[22:36:06] <gdamore> i would have thought sata would have the same setup issues, and just offered faster performance via a serial bus.
[22:36:19] <gdamore> but i'm ide/ata/sata ignorant.  know a tiny bit about scsi tho'.
[22:36:52] *** xzilla_ has joined #opensolaris
[22:37:01] <delewis> SB100 did not have DMA
[22:37:15] <delewis> it uses a shitty CMDK (?) chipset identical to the Ultra 10/Ultra 5
[22:37:20] <gdamore> i'm still not sure if i want to use the cobalt for this or not.  the alternative is to spend more $$ for a higher end Linkstation Pro or somesuch, and hack it for NFS.
[22:37:33] <delewis> (and actually what's in my Packard Bell router at the moment)
[22:37:35] <gdamore> i didn't know that.  no wonder sb100 sucked so badly.
[22:37:39] <delewis> :-)
[22:37:54] <delewis> and I'm fairly sure the SB150 uses the same chipset
[22:38:05] <delewis> basically, if it's Sun has IDE, stay way from it :-)
[22:38:16] <icon> delewis++
[22:38:33] <icon> im running SATA 2 in this opteron box, and ive been pretty pleased with it so far
[22:38:43] <gdamore> funny thing is, i think this raq2 has an LSI SCSI controller in it, but i'd have to solder on a header.  i would have thought cpu and network (10/100 tulip) would be the bigger bottlenecks
[22:38:47] <icon> even managed to find a compatible raid enclosure
[22:39:07] <icon> dec chips arent too bad
[22:39:09] <gdamore> the ide subsystem on my U20 occasionally hangs the whole system for a few seconds at a time.  it is really, really annoying.
[22:39:30] <gdamore> i know lots about the tulip. :-)  i wrote the afe driver for a tulip clone.
[22:39:31] <quasi> icon: who needs raid when you can use zfs? ;)
[22:39:40] <icon> quasi: true enough ;)
[22:39:45] <icon> i have yet to use hardware raid on this box
[22:39:49] <delewis> gdamore: are Tulips bad?
[22:39:52] <icon> but the enclosure is very nice to have
[22:39:56] <delewis> I was thought the DEC cards were pretty high-end for 10/100
[22:40:01] <delewis> (I've got in my router at the moment)
[22:40:08] <delewis> original DEC, too :-)
[22:40:11] <gdamore> pretty decent for 10/100, but they don't have a few features that are nice, like checksum offload
[22:40:16] <icon> quasi: http://arf.ubound.org/~sstallion/cruxis/
[22:40:22] <delewis> gdamore: how do they compare to 3coms?
[22:40:30] <icon> delewis: they dont really
[22:40:33] <gdamore> the 3coms are generally a bit better, more featureful.
[22:40:36] <icon> *nod*
[22:40:40] <delewis> ah, interesting.
[22:40:57] <icon> i have a couple of tulips laying around, but the 3coms still kick them
[22:41:02] <gdamore> the original tulips  are pretty good, but some of the clone devices are just plain awful.  the admtek chip is the best of the lot, though.
[22:41:13] <delewis> I've always had terrible experiences with the Linksys ones
[22:41:18] <delewis> but my DEC ones are fairly solid
[22:41:20] <icon> linksys was bad about the de's
[22:41:27] <gdamore> the linux tulip driver is just crap, though, because it has crap code to deal with oddball busted chips.
[22:41:34] <delewis> gdamore: ah
[22:41:45] <icon> i had decent perf on freebsd with a linksys de0
[22:41:53] <delewis> that would explain some of my odd experiences with Tulip chips (especially, the Linksys ones) on Linux
[22:41:56] <icon> i would regularly saturate a 10/100 duplex line (and stay sustained)
[22:41:59] <gdamore> the pnic was one of the worst clones ever.  really, really bad.
[22:42:04] <delewis> how's the Solaris Tulip driver fair?
[22:42:17] <gdamore> its got a lot of the same crap ancestry.
[22:42:21] <quasi> icon: the storage box looks like something I've seen before
[22:42:34] <icon> quasi: front-icon ?
[22:42:40] <icon> err: front-open ?
[22:42:51] <gdamore> i am a bit biased though.  my afe driver shucked all that shitty code, and started scratch.  i think i get close to hme performance.  it certainly can saturate a 10/100 line
[22:43:51] <gdamore> the only problem is that in order to stay safe for dynamic reconfiguration and avoid some edge cases, i use the cpu to copy data from descriptors, so it hits the cpu a bit harder than some other drivers do.  but i think modern cpus with smaller sizes (ethernet frames) copy is faster than dma setup/teardown
[22:44:43] <gdamore> btw, I am told afe is on-track for putback into Solaris soon.  Sun's QA found a definite race condition which I have fixed.
[22:45:17] * icon &
[22:45:51] <gdamore> &?
[22:46:02] <oxygene> gdamore: "detach"
[22:46:03] <Auralis> background, detached :)
[22:46:16] <andersmo> gdamore: think shell. =)
[22:46:41] <gdamore> ah.  sorry, mixing metaphors confused me. :-)
[22:47:17] <axisys> andersmo: thnx for your suggestion.. i check the config.log and found this http://www.rafb.net/paste/results/zCilGY17.html
[22:47:40] <axisys> then changed the configure command like this to fix it
[22:47:41] <axisys> LDFLAGS="-L/opt/csw/lib -Wl,-R/opt/csw/lib -lsocket -lnsl" ./configure --with-pcap-headers=/opt/csw/include
[22:47:49] <axisys> that did it
[22:48:21] <andersmo> axisys: yeah, there's a reason we like logs, isn't it? =)
[22:48:59] <axisys> andersmo: hell yeah
[22:51:52] <gdamore> heh.  my mail box is full from netbsd developers trying to figure out ftp.netbsd.org keeps crashing...
[23:08:42] <trygvis> icon: true, but I'm still lacking a windows machine :)
[23:32:57] *** crib has quit IRC
[23:33:40] *** crib has joined #opensolaris
[23:33:54] *** Disorganized1 has joined #OpenSolaris
[23:40:27] *** solaris-user has quit IRC
[23:42:16] *** Disorganized has quit IRC





top