July 22, 2011 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >
July 22, 2011 [00:13:04] *** DontKnwMuch has joined #openindiana[00:21:51] *** phretor has joined #openindiana[00:25:32] *** phretor_ has quit IRC[00:43:58] *** phretor_ has joined #openindiana[00:47:03] *** phretor has quit IRC[00:52:00] *** Alasdairrr is now known as AlasAway[01:06:10] *** magyar has joined #openindiana[01:19:19] *** movement has quit IRC[01:32:21] *** movement has joined #openindiana[01:32:57] *** winIO has quit IRC[01:56:31] *** sponix has joined #openindiana[02:01:13] *** InTheWings has quit IRC[02:43:38] *** bdrewery has quit IRC[02:45:54] *** jasonjwwilliams has joined #openindiana[02:47:26] <jasonjwwilliams> Is there a way to get "pkg info" to show the dependencies for a particular package?[02:50:07] *** dekar has quit IRC[02:53:21] <alanc> no, pkg contents does that[02:53:49] <alanc> pkg contents -t depend -o action.raw x11/server/xorg[02:53:51] <alanc> for instance[02:54:35] <jasonjwwilliams> alanc: Thank you that's exactly what I was looking for.[02:55:43] <richlowe> alanc: damn you[02:55:49] <Triskelios> the -o syntax is pretty obnoxious[02:55:50] <richlowe> I just got done emailing that :)[02:55:50] <alanc> ?[02:55:59] <richlowe> Triskelios: -o fmri is more natural[02:56:01] <alanc> heh, I just saw the e-mail[02:56:10] <richlowe> pkg contents -rHo fmri -t depend ...[02:56:15] <alanc> yeah, I just use action.raw for depend to show the type too[02:56:20] <jasonjwwilliams> Sorry guys. Got impatient.[02:56:39] <jasonjwwilliams> Writing a "package" provider for Chef for IPS[02:56:45] <alanc> what most people really want is probably: pkg contents -t depend -o fmri -a type=require[02:56:47] <richlowe> bdha: Hey, you had related stuff to this?[02:57:01] <richlowe> alanc: I don't think that's safe to assume[02:57:08] <richlowe> but I don't know how to || with -a[02:57:29] *** bdrewery has joined #openindiana[02:57:45] <richlowe> jasonjwwilliams: I'd expect if you need to query this stuff by hand you may have trouble.[02:57:54] *** dekar has joined #openindiana[02:58:08] <richlowe> You'd end up having to deal with variants and such yourself, and that's a completely painful mess.[02:58:37] <alanc> oh yeah, and action.raw would let you see if any dependencies are variant or facet specific[02:58:57] <jasonjwwilliams> richlowe: I'm using the apt "package" provider Chef already includes a template for what Chef expects a package provider to "provide". In this case, packages the current one is dependent on.[02:59:03] <richlowe> I ended up fixing the puppet provider by hacking out any semblence of intelligence it attempted to have.[02:59:28] <alanc> "pkg contents -t depend -o action.raw libx11" for instance showing that fmri=pkg:/x11/header/x11-protocols is only required if facet.devel=true[02:59:46] <richlowe> jasonjwwilliams: variants affect when an action within a package is actually used. The simplest case is variant.arch=sparc v. i386[02:59:54] <richlowe> jasonjwwilliams: so the commands we gave you may return dependencies that aren't actually operative.[03:00:04] <richlowe> jasonjwwilliams: some of which won't actually be satisfiable.[03:00:21] <bdha> richlowe: Nooo. I stay away from the package providers.[03:00:35] <richlowe> bdha: No, but you had done a bunch of other chefy/solaris-blah[03:00:45] <bdha> Yeah, I just use exec. :P[03:00:58] <richlowe> bdha: and this way I don't have to paste the salient bis of jasonjwwilliams's emails into IRC to you...[03:00:59] <richlowe> so.[03:01:01] <bdha> pkg has the same problem yum does.[03:01:18] <bdha> It iterates over each package, calling pkg/yum each time. Slow. Very slow.[03:01:34] <bdha> Passing a list is rather faster, but the providers don't do lists.[03:02:05] <bdha> Great, Apple has an "ARC" now too. :)[03:02:07] <richlowe> the puppet stuff does check installed state[03:02:20] <richlowe> though that's broken in the upstream puppet[03:02:27] <bdha> (Automatic Reference Counting)[03:03:44] <richlowe> bdha: I keep trying to convince the right pkg person that considering "running under chef or puppet" is a worthy use case to actually plan for.[03:03:48] <richlowe> thus far haven't succeeded[03:03:52] <dkeav> apple has an arc?[03:03:58] <richlowe> but I suspect parseable output, etc, for install will have the same sort of effect.[03:04:11] <bdha> dkeav: http://arstechnica.com/apple/reviews/2011/07/mac-os-x-10-7.ars/10[03:04:25] <richlowe> bdha: if you sysadmin types hadn't burned 'em out with ranting... ;)[03:04:45] <bdha> Yeah, it's obviously our fault.[03:04:56] <jasonjwwilliams> It doesn't look like the apt package provider in Chef does anything with the dependencies except when the pack in question is a virtual package.[03:05:05] <jasonjwwilliams> pack=package[03:05:06] <dkeav> ohhhh reference counting in obj-c[03:05:13] <dkeav> i thought you meant cache[03:05:14] <dkeav> heh[03:05:18] <bdha> jasonjwwilliams: Why are you trying to deal with depends in the provider? Why not let pkg(1M) do it?[03:05:32] <dkeav> like they shoehorned something like l2arc into HFS+[03:05:40] <bdha> dkeav: That would have been hilariously weird.[03:05:46] <dkeav> yes, yes it would[03:05:49] <dkeav> thats why i asked :D[03:06:16] <jasonjwwilliams> bdha: Not sure I have to deal with depends actually. Chef only seems to need that for systems that have a concept of virtual packages provided by other real packages…systems like apt.[03:06:47] <bdha> Metapackages? IPS has though. Consolidations. Or incorporations. Whatever they're called this week.[03:06:51] <bdha> s/though/those/[03:07:16] <bdha> jasonjwwilliams: Just not sure why you need to care. The logic should (probably) be constrained to:[03:07:20] <jasonjwwilliams> Virtual packages in APT are dummy packages that can be satisfied by one or more real packages.[03:07:29] <bdha> Is x installed? Is x the version I want? Okay, I'll install/upgrade x.[03:07:48] <bdha> Maybe I'm not thinking about it enough.[03:07:52] <jasonjwwilliams> In APT it allows a package A to depend on a virtual package, which can then be satisfied by either package X or Y or Z.[03:07:58] <bdha> Sure.[03:08:18] <bdha> But... why does the Chef provider care? Won't apt figure that out?[03:08:26] <bdha> Or does it ask which you want to install?[03:08:59] <jasonjwwilliams> bdha: It seems to specify the version number of the package if it's virtual.[03:09:09] <bdha> huh.[03:09:37] <jasonjwwilliams> bdha: But for IPS I don't think it'll matter. I'm more concerned about making sure my provider for IPS gives Chef all the info it needs for the node listing.[03:10:05] <bdha> ahh.[03:10:11] <bdha> That makes much more sense. Ok. :)[03:10:43] <jasonjwwilliams> bdha: Right now though trying to figure out if it's just the apt provider that gives a care about dependencies and virtual packages, or Chef actually grabs that data and puts it in the node attribute listings.[03:11:01] <bdha> Right.[03:12:01] *** master_of_master has quit IRC[03:12:07] <jasonjwwilliams> bdha: Doesn't look like it grabs that info.[03:13:32] <jasonjwwilliams> bdha: Found the reasoning for the apt provider's behavior: http://tickets.opscode.com/browse/CHEF-1439[03:14:00] *** master_of_master has joined #openindiana[03:15:32] <bdha> ahh. Yeah.[03:29:13] <richlowe> jasonjwwilliams, bdha: As best as I know, the way pkg does that wouldn't have similar issues.[03:29:29] <richlowe> but that gets you into dependency types, and exactly the kind of mess we were trying to save you from by suggesting just letting pkg figure it out.[03:30:31] <jasonjwwilliams> richlowe: I've ripped out the dependency/virtual packages logic. For now I'll let pkg figure it out as recommended. If it ends up being an issue later we can always correct it.[03:30:41] <alanc> yeah, since IPS doesn't really have a special type of metapackage, just packages only containing metadata[03:31:43] <jasonjwwilliams> My goal right now is to get Chef treating OpenIndiana as a first class citizen, so those that follow don't have to worry about it.[03:32:17] <jasonjwwilliams> Hence the other fixes I've submitted to Opscode in the past 72 hours.[03:32:48] <bdha> Yay![03:34:52] <richlowe> bdha: precisely why I wanted you to notice, weren't you doing similar things?[03:35:30] <jasonjwwilliams> bdha: You have a provider already submitted for iPS?[03:35:45] <richlowe> hah, no, bdha hates packages :)[03:35:59] <jasonjwwilliams> LOL.[03:36:45] <dkeav> hate is such a strong word[03:36:49] <dkeav> loathe?[03:37:23] <bdha> Avoid.[03:37:42] <bdha> richlowe: No, holoway wanted to get the full stack Chef installer working on Solaris, so I helped with that.[03:37:49] <richlowe> bdha: Ah[03:37:54] <bdha> I haven't written any custom Chef code for Solaris.[03:37:56] *** magyar has quit IRC[03:38:12] <bdha> It needs zones and ZFS primitives. It also needs better SMF support.[03:38:20] <bdha> And as jasonjwwilliams obviously knows, better packaging support.[03:38:32] <bdha> btw, the full stack installer uses fpm to spit out SRV4 or ipkg. :)[03:38:43] <richlowe> point me at that?[03:40:37] <bdha> https://github.com/adamhjk/omnibus[03:40:44] <bdha> omnibus is pretty badass in and of itself.[03:40:56] <bdha> omnibus is the building framework.[03:41:33] <bdha> oi_148 is down near the bottom. :)[03:50:42] <bdha> http://arstechnica.com/tech-policy/news/2011/07/a-pound-of-flesh-how-ciscos-unmitigated-gall-derailed-one-mans-life.ars[04:04:02] *** kdavy_ has joined #openindiana[04:04:12] *** DrLou has quit IRC[04:07:09] <sivanov__> so, given a chance spend $40 on IBM BR10i (lsisas 1068e) or ibm m1015 (lsi 9210-8i) what to buy ?[04:07:33] <sivanov__> if pci-e is only 1.0[04:47:12] <jasonjwwilliams> My experience with 1068s in SunFire's has been hit and miss.[04:48:41] <jasonjwwilliams> I don't believe the 9210 is supported but the MPT SAS driver but folks seem to have good luck with LSI's Solaris drivers for it.[04:54:50] *** mnaser has joined #openindiana[05:05:44] *** jasonjwwilliams has quit IRC[05:13:32] *** skeeziks1 has joined #openindiana[05:21:03] *** blues has quit IRC[05:25:48] *** Naresh has quit IRC[05:28:02] <Botanic> how can I change the port for snmpd? The config file doesnt look the same as normal...[05:31:40] <baitisj> hey guys, any ideas how to disable routing through an interface at reboot? my hostname.hme0 has "192.168.2.3 -router" but still comes up w/router flag[05:32:00] <baitisj> i have to ifconfig hme0 -router every time :-([05:38:14] *** syoyo_ has joined #openindiana[05:42:17] *** redgone has joined #openindiana[05:45:06] *** POloser has joined #openindiana[05:48:33] *** Thrae has quit IRC[05:49:07] *** Thrae has joined #openindiana[05:49:18] *** syoyo_ has quit IRC[05:49:46] *** kart_ has joined #openindiana[06:01:59] *** redgone has quit IRC[06:05:17] *** POloser has left #openindiana[06:11:45] *** POloser has joined #openindiana[06:23:43] *** Thrae has quit IRC[06:24:16] *** Thrae has joined #openindiana[06:44:21] *** flyz_ has joined #openindiana[06:46:22] *** flyz has quit IRC[06:55:04] <madwizard> Coffee[07:00:46] *** syoyo_ has joined #openindiana[07:21:56] *** syoyo_ has quit IRC[07:23:17] <baitisj> do more people use opencsw or blastwave these days?[07:23:34] <baitisj> ("what's better?")[07:24:17] <bdha> The pkgsrc IPS repo may be useful to you.[07:24:41] <bdha> https://www.illumos.org/projects/worsoe[07:26:01] <richlowe> My understanding was "SFE or pkgsrc"[07:26:05] <richlowe> 'cos those folks are less shouty[07:26:37] <sivanov__> is there any built-in tools to diagnose that a SATA disk is about to die?[07:27:00] <sivanov__> i get 100%b on 3 disks too often...[07:27:03] <richlowe> if you mean SMART stuff, no.[07:27:08] <richlowe> iostat has error counters.[07:27:11] <richlowe> ZFS notices data badness.[07:27:28] <richlowe> never entirely sure about the state of FMA and disks[07:27:39] <bdha> Yeah, it's dubious.[07:27:42] *** kdavy_ has quit IRC[07:28:26] <sivanov__> the two disks of them are old 750Gb WD green[07:31:26] <baitisj> sivanov__, interesting you asked this. I actually updated my pkgutil sources because I want to get smartmontools installed :-)[07:31:38] *** descipher has joined #openindiana[07:32:05] <baitisj> great link, richlowe .. thanks[07:32:19] <bdha> richlowe: Man, taking my props.[07:32:38] *** descipher_ has quit IRC[07:32:44] <sivanov__> the disks sit on same vdev and while remaining three disks receive 45-55% read load the two get 98-100% load[07:33:17] <Loafamotive> sivanov__: you can build smartmontools[07:33:23] <sivanov__> i know[07:34:02] <baitisj> i like having a package manager looking over my shoulder when I install stuff[07:34:17] <baitisj> file conflicts / overwrites bad[07:34:24] <baitisj> and i'm too lazy to check every makefile[07:34:25] <bdha> PREFIX.[07:34:32] <bdha> It's a thing. Use it.[07:34:39] <baitisj> I've seen broken packages ignore PREFIX[07:34:46] <Loafamotive> heh, you must not have been around when package managers used to cause more problem than they solved[07:34:49] <Loafamotive> ;)[07:34:55] * bdha hasn't. At least not for popular things.[07:35:05] <bdha> Loafamotive: ...used to?[07:35:10] <Loafamotive> haha[07:35:11] <herzen> bdha: I don't believe a pkgsrc IPS repo exists[07:35:18] <bdha> herzen: It does.[07:35:29] <sivanov__> interesting that 3 disks make 12000kr/s and two disks do 5500kr/s[07:35:32] <bdha> Or I'm pretty sure it does.[07:35:42] <baitisj> true, I haven't really experienced the wrath of package managers.[07:35:49] <baitisj> I kind of view them as a prerequisite.[07:36:04] <sivanov__> i thought that the IO load on read should me more speaded around disks in 1 vdev[07:36:07] <baitisj> if some distro has extremely poor package management, i shun it.[07:36:09] <sivanov__> *raidz vdev[07:36:14] *** hsp has joined #openindiana[07:37:05] <bdha> You could also set PKG_PATH appropriately: http://pkgsrc-repo.uk.openindiana.org/[07:37:15] <bdha> But I'm almost positive I saw a dev IPS repo for it.[07:37:25] <herzen> bdha: what's the URL?[07:37:52] <sivanov__> and now i get miserable speeds reading big data sizes with iozone from 4GB FC:[07:37:53] *** Botanic has quit IRC[07:37:59] <sivanov__> Children see throughput for 4 readers = 136635.53 KB/sec[07:38:08] <sivanov__> Children see throughput for 4 re-readers = 98641.03 KB/sec[07:38:33] <sivanov__> from 15 disks in 3 raidz vdevs[07:38:35] <bdha> herzen: I can't find it. It wasn't published and my google-fu is failing me. :([07:38:43] <bdha> worsoe or AlasAway would know.[07:38:56] <baitisj> sivanov__, maybe dd > /dev/null to see if one disk is significantly slower than the others?[07:39:30] <herzen> last I saw worsoe talk about this, he said he doesn't have time to create an IPS repo[07:39:49] <bdha> herzen: I really thought someone from OI had, though.[07:40:02] <bdha> baitisj: You could get that from iostat as well. :)[07:40:11] *** echobinary1 has quit IRC[07:41:19] <herzen> the problem is that pkgsrc is a BSD packaging system that hasn't really been ported to Solaris/OI[07:41:41] <bdha> Alasdair (or someone) wrote a pkgsrc -> ipkg converter.[07:41:50] <bdha> There was already a pkgsrc -> SRV4 tool _in_ pkgsrc.[07:42:01] <bdha> It's entirely possible I'm being crazy, but I don't think so. I just can't find it.[07:42:02] <baitisj> bdha... I thought that zfs dynamically balances reads based on drive performance, therefore reads issues would be "balanced out" ??[07:42:24] <baitisj> thought that the only real way to get a metric on a drive was to access it directly.[07:42:30] <bdha> baitisj: You'd still see latency issues (a_svct).[07:42:48] <bdha> I don't know how well ZFS's read balancing works, if at all.[07:42:49] <richlowe> As best as I recall the conversation, it balances reads based on balance[07:42:56] <richlowe> in the sense of "putting shit on scales"[07:43:00] <richlowe> or "that scene in die hard 3"[07:43:02] <richlowe> nothing about historic perf.[07:43:29] <bdha> richlowe: Am I being crazy about the IPS IPP repo?[07:45:23] <herzen> Loafamotive: have you gotten smartmontools to run on OI? I can build it, but it can't do anything useful, as I recall.[07:46:06] <Loafamotive> i have to admit I havnt run it on the OI kernel[07:46:31] * Loafamotive bows head in shame[07:49:52] <richlowe> bdha: I don't know.[07:51:52] *** flyz_ has quit IRC[07:52:32] *** McBofh has quit IRC[07:52:38] <bdha> bah.[07:53:19] *** ThomasB2k has quit IRC[07:54:05] *** ThomasB2k has joined #openindiana[07:54:10] *** trn has quit IRC[07:55:12] *** McBofh has joined #openindiana[07:56:48] *** flyz has joined #openindiana[07:57:29] *** mnaser has quit IRC[07:58:34] *** trn has joined #openindiana[08:05:42] *** gea has joined #openindiana[08:08:19] *** bdha has quit IRC[08:09:55] <sivanov__> herzen, you need to use -d sat,12 with smartctl most likey[08:11:25] *** dekar has quit IRC[08:16:29] *** dekar has joined #openindiana[08:16:34] *** bdha has joined #openindiana[08:18:10] <baitisj> yep[08:18:22] <baitisj> i confirm that -d sat,12 is required on my machine[08:18:24] <baitisj> just figured that out[08:18:32] <baitisj> sat,16 does not work[08:20:34] <herzen> sivanov_: thanks for the suggestion. I'll try that some time[08:21:45] <herzen> sivanov__, I mean[08:25:19] <sivanov__> its very strange, i made four zvols, blocksize=64k. exported them to FC and got read speed 50-90MB/sec over FC[08:25:44] <sivanov__> locally i get 350MB/s[08:25:49] *** gea has quit IRC[08:30:30] *** McBofh has quit IRC[08:32:27] *** lkocman has joined #openindiana[08:32:48] *** bens1 has joined #openindiana[08:34:12] *** kart_ has quit IRC[08:34:42] *** kart_ has joined #openindiana[08:36:36] *** McBofh has joined #openindiana[08:41:14] *** McBofh has quit IRC[08:41:49] *** McBofh has joined #openindiana[08:46:59] *** McBofh has quit IRC[08:47:47] *** McBofh has joined #openindiana[08:48:48] *** macky22 has joined #openindiana[08:52:32] <macky22> cannot install. system always freezes at "preparing ... image ..." . What can i do find the reason ?[08:54:57] *** gea has joined #openindiana[08:55:02] *** McBofh has quit IRC[08:56:18] *** McBofh has joined #openindiana[08:56:20] *** anikin has joined #openindiana[09:00:01] *** gea has quit IRC[09:00:09] *** skeeziks1 has quit IRC[09:02:42] *** McBofh has quit IRC[09:03:14] *** McBofh has joined #openindiana[09:12:17] <macky22> does happen on Live GUI, Text and Install Images. Any hints how i can investigate for the reason ?[09:12:42] <sivanov__> are you installing to USB?[09:12:55] <richlowe> boot with -kvd added to the kernel$ line (hit 'e' in grub)[09:13:01] <richlowe> it'll drop to the debugger[09:13:04] <richlowe> snooping/W 1[09:13:04] <sivanov__> flash disk[09:13:10] <richlowe> moddebug/W 80000000[09:13:11] <richlowe> :c[09:13:32] <sivanov__> i had issues doing so with s11e[09:13:49] *** McBofh has quit IRC[09:15:17] *** McBofh has joined #openindiana[09:15:42] <richlowe> it'll be verbose about modules, and enable the deadman timer, it may or may not fire after the system has been hung for a minute[09:15:42] <richlowe> I grow more and more convinced it's broken by the day.[09:15:42] <richlowe> but it's worth a try[09:20:08] *** hsp has quit IRC[09:23:01] *** |AbsyntH| has joined #openindiana[09:27:44] *** Naresh has joined #openindiana[09:28:11] *** sudosu has joined #openindiana[09:32:59] *** McBofh has quit IRC[09:34:15] *** McBofh has joined #openindiana[09:47:59] *** movement has quit IRC[09:51:00] *** bens1 has quit IRC[09:51:11] *** McBofh has quit IRC[09:52:00] *** McBofh has joined #openindiana[09:57:31] *** McBofh has quit IRC[09:58:47] *** McBofh has joined #openindiana[10:00:34] *** movement has joined #openindiana[10:04:25] *** McBofh has quit IRC[10:04:34] *** flyz has quit IRC[10:04:38] *** Worsoe has joined #openindiana[10:11:42] *** McBofh has joined #openindiana[10:23:33] *** Whoopsie has joined #openindiana[10:23:33] *** ChanServ sets mode: +v Whoopsie[10:25:06] *** McBofh has quit IRC[10:25:49] *** tsoome has quit IRC[10:36:30] <macky22> thanks for your feedback. I noted this down and will test as soon as possible.[10:39:17] <macky22> sivanov__: i was trying the installation/livecd from a USB CD Drive. One time i got LiveCD going into console, but this did never happen again :([10:41:13] <sivanov__> omg, i wonder if anybody sells supercap backups for hard disks[10:46:43] <macky22> not sure what you mean. UPS, BBU, .. ?[10:47:10] <sivanov__> i think bbu[10:48:34] <sivanov__> separate devices for those who cannot afford enterprise ssd[10:51:08] <macky22> BBU is the Battery Backup Unit used by RAID Controllers, i think same thing as the ZIL in ZFS.[10:53:26] <macky22> ? not sure how to compare this to an SSD.[10:53:47] *** tsoome has joined #openindiana[10:53:51] <macky22> Can i disable the ZIL if i have a BBU on my Controller ?[10:54:45] <AlasAway> macky22: the zil isn't about battery backup or not, it's about synchronous writes[10:54:55] <AlasAway> if you turn it off, any synchronous write will become an async one[10:55:02] <AlasAway> regardless of a bbu[10:56:30] <macky22> ok i think i messed it up though.[10:57:08] <tsoome> zip protects you from server crash/power failure, same way as battery protects your controller[10:57:25] <tsoome> s/sip/zil/[10:57:29] *** bens1 has joined #openindiana[10:57:29] <tsoome> nah[10:57:31] <tsoome> :D[10:58:47] <tsoome> it is possible to disable it (zfs set sync=) but thats not something you wanna do without thinking hard first;)[11:03:27] <sivanov__> tsoome, today i'me made 64k recorsize zvols (were 128k ones) and get 50% slower read speed on FC and about same local speed[11:03:44] *** Micr0mega has joined #openindiana[11:03:49] <sivanov__> i think, maybe there is a bug somewhere...[11:04:31] <tsoome> with 64k recordsize, you need to do 2x more iops to read same amount of data[11:04:52] <tsoome> assuming you are reading large chunks[11:04:56] <sivanov__> but locally i get same decent speed[11:06:35] <sivanov__> and i dont think that qla2460 is iops limited @1600 iops (i get less than 100MB/s)[11:06:50] <sivanov__> and locally i get 400-500[11:07:20] *** lblume has joined #openindiana[11:10:47] <tsoome> what I really haven't figured out is how all those different block sizes do add up - if you export un with costar, your lun physical block size is 512B; you can set block size with stmfadm create-lu, but that will set logical block size. now if you get single 512B write from comstar client, that would have to be placed into volrecordsize and done with sync write (by default). now to make things even more "fun", if you have multiple vdevs, that IO[11:10:47] <tsoome> will be striped across vdevs (how exactly?) I think there is a lot of ground for optimizations...[11:11:19] <macky22> strange thing ! maybe FC related ?, but can't imageine FC transfers in about 2k Chunks. Any possibility to try with iSCSI ?[11:13:28] *** ThomasB2k has quit IRC[11:14:16] <macky22> is COMSTAR involved there or is it only for FCoE/iSCSI ?[11:15:27] <tsoome> comstar does FC/FCoE/iSCSI/...[11:15:54] <macky22> tsoome: you can really confuse me (related to your blocksize thoughts)[11:16:20] <tsoome> the comstar itself is not the issue as sivanov did test with ramdisk and got nice results[11:17:08] <tsoome> its how the zfs layer will kick in[11:17:09] *** McBofh has joined #openindiana[11:17:09] <sivanov__> or when results are cached[11:17:56] <tsoome> but then again, zil/slog can't possibly affect the reads[11:18:29] <macky22> haven't researched about COMSTAR/FC, i would have thought this is not involved at all when reading locally ![11:18:39] <macky22> and required when doung FC / iSCSI ![11:18:43] <sivanov__> and pre-reading large data does not work everytime[11:18:45] <tsoome> sure thing[11:19:24] <sivanov__> ramdisk<->fc nice speed[11:19:34] <sivanov__> cache<->fc nice speed[11:19:37] <tsoome> also local writes are all async by default (unless your test app is using O_SYNC, or if you zfs set sync=always)[11:19:48] <sivanov__> zfs<->local slow speed[11:20:00] <sivanov__> zfs<->local high speed[11:20:02] <sivanov__> *[11:20:18] <sivanov__> zfs<->fc low seed[11:20:25] <sivanov__> macky22, like that[11:21:05] <sivanov__> tsoome, i get OK write speed over FC[11:21:29] <sivanov__> 270-280 of 400[11:21:58] <sivanov__> as the luns are created with wcd=false[11:22:19] <tsoome> aye, well then you don't need slog;)[11:23:33] <sivanov__> right now its doing random write test @64k block and zpool iostat shows some 250MB/s average[11:24:01] <tsoome> compare that spool iostat with iostat from client side[11:24:09] <tsoome> zpool*[11:25:01] <sivanov__> client is freebsd, using FC devices as raw scsi disks[11:25:10] <sivanov__> *FC luns[11:25:30] <tsoome> freebsd should have iostat as well i think[11:25:40] *** McBofh has quit IRC[11:26:31] <macky22> ok. can't help anyway ![11:27:22] <sivanov__> well, the four devices on client are doing 50-60mb/s each[11:27:50] <sivanov__> Children see throughput for 4 random writers = 183842.37 KB/sec[11:28:16] *** McBofh has joined #openindiana[11:28:29] <tsoome> mb as megabit or byte?:P[11:28:36] <sivanov__> byte[11:28:43] <tsoome> ~200MB/s[11:28:47] <tsoome> as total[11:29:23] <sivanov__> so about 3200 write iops[11:29:59] <tsoome> thats not that bad at all[11:30:15] <tsoome> how many physical disks in pool?[11:30:20] <sivanov__> 15[11:30:24] <tsoome> sata?[11:30:28] <sivanov__> yes[11:31:02] <sivanov__> also WD Green :)[11:32:24] <sivanov__> two things aer great about them: low temperature (so i can keep rpms of fans low) and 4W consumption when idle[11:32:45] <tsoome> roughly 213 IOPS per disk - the IOPS you get is delivered by the help of the cache, and not raw disk performance, i guess the disks in pool didn't show up that many iops;)[11:33:48] <sivanov__> actually IIRC the disks do ~270KB/s @4kb[11:33:54] *** McBofh has quit IRC[11:34:06] <sivanov__> or was it ~500KB/s[11:34:10] <sivanov__> dont remember[11:35:24] <sivanov__> what may cause problems is that i have different size vdevs[11:35:33] <sivanov__> 2TB*5 vdev[11:35:40] <sivanov__> 1TB*5 vdev[11:35:47] *** nikolam has joined #openindiana[11:35:49] <sivanov__> 750GB*5 vdev[11:36:30] <tsoome> it will cause eventually, but till the smaller vdevs are not files, that shouldn't really matter[11:36:35] <tsoome> filled*[11:36:37] <sivanov__> i've done some copying around to balance data accros them[11:36:53] <sivanov__> but now the pool is 65% full[11:48:09] *** McBofh has joined #openindiana[11:51:22] *** Agony has joined #openindiana[11:51:24] *** held has quit IRC[12:04:47] *** McBofh has quit IRC[12:07:02] *** |AbsyntH| has quit IRC[12:13:50] *** Whoopsie has quit IRC[12:17:20] <macky22> need to leave now. bye.[12:17:30] *** macky22 has quit IRC[12:26:40] *** held has joined #openindiana[12:46:45] *** hsp has joined #openindiana[12:51:21] *** tsoome has quit IRC[13:14:28] *** tsoome has joined #openindiana[13:19:49] *** movement has quit IRC[13:21:53] <nettezzaumana> anyone having access to t3+ by luck ?[13:26:24] *** spanglywires has joined #openindiana[13:26:39] *** spanglywires has quit IRC[13:33:46] *** movement has joined #openindiana[13:34:34] *** AlasAway is now known as Alasdairrr[13:37:24] *** madevil has joined #openindiana[13:37:33] <madevil> hello![13:38:59] <madevil> have anyone tried oi_151 on a laptop? I did, gets soooo hot... perhaps laks cpufreq hal stuff like explained here http://hub.opensolaris.org/bin/view/Community+Group+laptop/Cpufreq ... am I wrong? please someone just slap me out of my ignorance!!![13:40:23] <madevil> anyone please... am I just alone? hasn't noone tried it on a laptop?[13:40:44] <madevil> intel... c2d? anyone...????? please... come on...[13:41:35] *** ThomasB2k has joined #openindiana[13:42:15] *** DrLou has joined #openindiana[13:42:16] *** ChanServ sets mode: +o DrLou[13:42:28] <madevil> freebsd is so cool on the laptop, iron solid ZFS port... but then I got no support for my sis 0x6351 chipset... which is supported with sismedia on linux side... so I'm on ubuntu now... but looking for a 0.000000009th reason to go back to oi....[13:43:37] <tsoome> sorry, using lion :P[13:43:40] <tomww> what CPU type? old AMD (Turion) get fried easily[13:43:43] <madevil> still there is no sismedia on either oi or freebsd, but at least oi has a rather higher res than freebsd when it comes to vesa video...[13:45:43] <madevil> ok.. my desktop is on 10.6.8... snow leo... just fine... but my laptop sucks on osx... freebsd rocks on it... but no video driver... and 800x600 is not good enough... oi gets good res... but too hot 'coz no cpufreq hal routines I guess... opensolaris has a page on it... I don't get to understand the final steps... so... is there no cpufreq on oi?[13:46:41] <Alasdairrr> madevil: there is plenty of power management stuff[13:46:43] <Alasdairrr> you just have to turn it on[13:46:45] <madevil> just a simple question... come on.. is there no one else here to answer it?[13:47:03] <Alasdairrr> its all in the power config file in /etc[13:47:24] <Alasdairrr> man power.conf[13:48:02] <madevil> well... I went ahead and installed oi... battery seemed to work just fine... ok... I'll take a look... but still... is there cpufreq stuff in it? can you say it for sure?[13:48:13] <tsoome> man google[13:48:44] <madevil> *sigh* right... go ahead and try... all you'll find is this page from opensolaris...[13:48:49] <madevil> http://hub.opensolaris.org/bin/view/Community+Group+laptop/Cpufreq[13:49:11] <madevil> be my guest, I'll buy you a beer if you find ANYTHING about it...[13:49:57] <tsoome> This work has been integrated to nevada build 91. so it should be there:P[13:51:07] <madevil> ok.. right... i'll fire my vmware... open oi_151 livecd... be right back...[13:52:52] <madevil> why would it get so hot then?[13:53:06] <tsoome> clean the dust?[13:53:09] *** POloser has left #openindiana[13:53:34] * tsoome just did finish cleaning damn laptop[13:54:33] <madevil> what? so freebsd didn't mind the dust but oi does...? that's the sthangest thing someone ever said about such stuff... lol... you are rather funny...[13:55:51] <tsoome> well, keep in mind the solaris has never designed for desktops, altho it got some power management, its really quite basic compared to the needs of laptops etc.[13:56:54] <madevil> alright, I'm on vm-oi... just fired up terminal and did man power.conf... there is some cpu-pm stuff... I'm reading it...[13:57:55] <tsoome> it can also be some specific hardware support issue - my last attempt to use OI on W2100z did prove bad because of that - the "idle" system was still quite loaded and system had to turn on vents....[13:58:10] <madevil> oooh man... isn't there just a [*]Activate CPU-PM stuff?[13:58:47] <tsoome> what that should do and why?[14:00:17] <madevil> keep the cpu at lowest hz while idle so it remains cool and then bring it up to hightes when loaded... like on-demand? I don't think is so hard to understand... I don't beleive one would get the wrong impressions when talking about this.... is just quite simple to understand...[14:00:43] <Micr0mega> madevil: http://tinyurl.com/4y9uv68[14:01:15] <tsoome> just get some real desktop os…. :P[14:01:40] <madevil> lol[14:02:21] <madevil> that's quite fu**** up in the head...[14:02:23] <madevil> lol[14:02:35] <madevil> maaan... thanks anyone.. everyone... you all...[14:02:35] <gerard13> hello all, http://dlc.sun.com/opensourcecode/solaris/sol-11-exp-201011-GPLSource_1.zip does this archive permit to upgrade OI to 151 ?[14:03:42] <gerard13> oups sorry, i didn't notice the date[14:04:46] <madevil> Micr0mega.... how did you do that?[14:05:05] <madevil> nevermind... I'm just teasing... thanks...[14:06:12] *** Spunkmaster has joined #openindiana[14:09:44] *** eto has quit IRC[14:11:32] *** eto has joined #openindiana[14:11:43] <madevil> I was rather pesimistic while googling... I feel like such a retard... thanks tsoome, I owe you an apology...[14:11:53] <tsoome> ?:P[14:12:28] <tsoome> So what was it?[14:12:29] <madevil> I mean, you were right, Micr0mega pointed the hint... thanks both...[14:12:41] <madevil> http://tinyurl.com/4y9uv68 lol[14:12:52] <tsoome> ;)[14:12:56] <Micr0mega> :)[14:13:01] <madevil> there is a bunch!!![14:13:48] *** |AbsyntH| has joined #openindiana[14:13:49] <tsoome> umm…. why such thing needs to be configured in first place:P[14:15:19] <madevil> that's exactly what I thought... pardon my english... I'm from vnzla by the way... anyhow... why??? just why!? *whimps* buaaahhhh....[14:17:18] <madevil> man, I hate ubuntu...[14:17:51] <tsoome> well, probably same reason why they set system defaults in ~/.profile ...[14:18:01] <madevil> I'm out, forgive me for being.... nevermind...[14:18:08] *** madevil has quit IRC[14:20:10] <Micr0mega> well, that was interesting :P[14:20:29] *** Edgeman has quit IRC[14:20:58] *** Spunkmaster has quit IRC[14:21:03] *** Spunkmaster has joined #openindiana[14:21:56] <tsoome> well, thats why sane defaults are important...[14:23:03] <Micr0mega> to keep the uhm, less sane people out? :P[14:24:13] <Micr0mega> nah just joking, I thought he was quite funny[14:24:15] <tsoome> thats one way, sure[14:32:56] *** DontKnwMuch_ has joined #openindiana[14:34:05] <DontKnwMuch_> hi :) I am installing my third oi system, and what I have is ifconfig telling me "bad address" when I try to plumb the existing (which was not there at install) nic...[14:34:49] <DontKnwMuch_> and I already thought I finally learned some stuff in the weeks passed :([14:35:59] <gerard13> perhaps the new nic driver isn't loaded?[14:36:41] <DontKnwMuch_> it is dladm show-link shows it[14:36:48] <DontKnwMuch_> e1000g0[14:37:15] <DontKnwMuch_> and rge0 which I do not want to use anyway[14:37:55] <gerard13> did you unplumb rge0 ?[14:40:23] *** Alasdairrr is now known as AlasAway[14:40:25] <DontKnwMuch_> no, it was disabled at install in bios, so it also is in the same state as e1000g0[14:40:54] <lblume> DontKnwMuch_: Sounds more like you bungled the command line[14:40:59] <DontKnwMuch_> I can not plumb or unplumb any of them[14:41:50] <lblume> You get that messge when you exchange device name and command[14:42:00] <lblume> it should be: ifconfig <nic> plumb[14:42:08] <DontKnwMuch_> ifconfig e1000g0 plumb[14:42:33] <lblume> Not ifconfig plumb <nic>[14:42:35] <DontKnwMuch_> crap[14:42:56] <DontKnwMuch_> I am on something or what :) thanks ... argh....[14:43:12] <lblume> There's a convoluted man for that :-)[14:43:36] <DontKnwMuch_> :)[14:46:39] <gerard13> great lblume![14:48:31] *** Edgeman has joined #openindiana[14:48:44] <lblume> Reemmbering my own mistakes is not that difficult :-P[14:53:52] *** tsukasa` has joined #openindiana[14:57:18] *** tsukasa has quit IRC[15:04:47] *** hsp has quit IRC[15:05:59] *** ekix has quit IRC[15:09:14] *** lkocman has quit IRC[15:09:44] *** hsp has joined #openindiana[15:09:53] *** kytibe has quit IRC[15:10:14] *** Spunkmaster has quit IRC[15:12:19] *** kytibe has joined #openindiana[15:14:53] *** ekix has joined #openindiana[15:16:20] *** mnaser has joined #openindiana[15:17:44] *** ChanServ sets mode: +o bdha[15:25:41] *** Worsoe has quit IRC[15:34:48] *** gea has joined #openindiana[15:43:07] *** jedi4ever has joined #openindiana[15:44:46] *** gea has quit IRC[15:49:25] *** mnaser has quit IRC[15:50:20] *** jedi4ever has quit IRC[15:50:51] *** sudosu has quit IRC[15:51:50] *** tsoome has quit IRC[15:56:25] *** mnaser has joined #openindiana[15:58:32] *** dkeav has quit IRC[16:04:10] *** sudosu has joined #openindiana[16:10:27] *** nikolam has quit IRC[16:10:31] *** gea has joined #openindiana[16:13:03] <RoyK> wtf. bomb blast in central oslo...[16:16:02] <viridari> Mitt hjerte er i Oslo i morges.[16:16:58] <quasi> prolly someone with the car full of illegal booze trying to light a cigarette[16:17:36] <viridari> unlikely considering the elevation of the explosion[16:18:07] <viridari> the news here is saying it looks like the explosion may have started in or adjacent to the PM's office[16:18:16] <viridari> but they confirm the PM was not in his office and is fine[16:18:25] <viridari> other people weren't so lucky[16:19:15] *** baitisj has quit IRC[16:21:55] *** kdavy_ has joined #openindiana[16:22:41] *** anikin has quit IRC[16:23:30] *** anikin has joined #openindiana[16:25:18] *** McBofh has joined #openindiana[16:29:26] *** elgar has joined #openindiana[16:36:30] <RoyK> quasi: it'll take a BIT more than that to blow out windows in a 500m radius or so[16:37:33] <joffe> it's definitely not an accident[16:37:36] <joffe> the blast was too big[16:37:39] <joffe> well multiple blasts[16:37:52] <joffe> and an auspicious location for an 'accident' too - govt quarters, PM office[16:40:30] <nettezzaumana> pls help .. what the F*CK[16:40:31] <nettezzaumana> # /usr/bin/env python; file /usr/bin/env; pkginfo | grep python27[16:40:31] <nettezzaumana> /usr/bin/env: No such file or directory[16:40:31] <nettezzaumana> /usr/bin/env: ELF 32-bit LSB executable 80386 Version 1, dynamically linked, stripped[16:40:34] <nettezzaumana> application BWpython27 python - A high-level scripting language.[16:41:37] <alanc> presumably whatever the hell BWpython27 is didn't put a program named python in your $PATH[16:41:53] <alanc> that's the error env gives when it can't find the program you told it to run[16:42:36] *** gea has quit IRC[16:45:21] <nettezzaumana> alanc: oh really, i'm reatrded .. sorry, friday[16:46:39] <alanc> alanc@also:/tmp/pkg-test [7:45am - 493] env nosuchfile[16:46:39] <alanc> env: No such file or directory[16:46:39] <alanc> alanc@also:/tmp/pkg-test [7:45am - 494] env date[16:46:39] <alanc> Friday, July 22, 2011 07:50:49 AM PDT[16:49:35] *** tsoome has joined #openindiana[16:50:27] <nettezzaumana> alanc: my apologies .. i never use env shit .. it was in some script .. i just use usr/bin/whatever but never that .. i unfortunatelly uninstalled on this host SUNW python and forget that .. all other hosts were okay but this one contains BWpython27 and complained[16:50:58] <nettezzaumana> alanc: i asked just before i used a brain on it[16:51:18] <nettezzaumana> **forgot[16:53:41] *** McBofh has quit IRC[17:03:37] *** McBofh has joined #openindiana[17:06:30] *** ivwang has joined #openindiana[17:24:07] <bdha> http://www.linux-kvm.org/page/KVM_Forum_2011 # ^F cantrill[17:29:24] *** miine has quit IRC[17:35:58] *** miine has joined #openindiana[17:37:36] *** Micr0mega has left #openindiana[17:46:28] *** AlasAway is now known as Alasdairrr[17:47:58] *** McBofh has quit IRC[17:48:12] *** |AbsyntH| has quit IRC[17:49:49] *** McBofh has joined #openindiana[18:02:55] *** Agony has quit IRC[18:08:42] *** McBofh has quit IRC[18:08:50] *** DontKnwMuch has quit IRC[18:10:01] *** alanc has quit IRC[18:11:06] *** McBofh has joined #openindiana[18:11:17] *** tomww has quit IRC[18:17:02] *** alanc has joined #openindiana[18:17:02] *** ChanServ sets mode: +o alanc[18:18:37] *** McBofh has quit IRC[18:21:22] *** McBofh has joined #openindiana[18:22:41] *** laserbled has joined #openindiana[18:43:12] *** anikin has quit IRC[18:43:36] *** Naresh has quit IRC[18:46:04] *** tomww has joined #openindiana[18:46:04] *** ChanServ sets mode: +o tomww[18:46:50] *** McBofh has quit IRC[18:47:41] *** McBofh has joined #openindiana[18:50:12] *** ivwang has quit IRC[18:50:32] *** ivwang has joined #openindiana[18:50:42] *** ivwang has quit IRC[18:53:21] *** McBofh has quit IRC[18:54:01] *** McBofh has joined #openindiana[19:04:28] *** McBofh has quit IRC[19:06:17] *** McBofh has joined #openindiana[19:06:54] *** InTheWings has joined #openindiana[19:09:51] *** DeanoC has quit IRC[19:14:31] *** McBofh has quit IRC[19:18:32] *** McBofh has joined #openindiana[19:24:28] *** McBofh has quit IRC[19:25:02] *** McBofh has joined #openindiana[19:28:43] *** Botanic has joined #openindiana[19:36:02] *** Botanic has quit IRC[19:37:46] *** McBofh has quit IRC[19:40:27] *** yalu has quit IRC[19:40:29] *** McBofh has joined #openindiana[19:43:50] *** held has quit IRC[19:44:27] *** yalu has joined #openindiana[19:46:15] *** McBofh has quit IRC[19:48:33] *** DeanoC has joined #openindiana[19:48:33] *** ChanServ sets mode: +o DeanoC[19:49:39] *** dkeav has joined #openindiana[19:50:57] *** McBofh has joined #openindiana[19:54:03] *** gea has joined #openindiana[19:54:43] *** paularmstrong has joined #openindiana[19:56:13] *** dimonov has quit IRC[20:03:38] *** wonslung has joined #openindiana[20:03:43] *** Botanic has joined #openindiana[20:04:50] *** paularmstrong has quit IRC[20:07:16] *** raichoo has joined #openindiana[20:11:10] *** kdavy_ has quit IRC[20:13:55] *** held has joined #openindiana[20:22:06] *** Botanic has quit IRC[20:39:51] *** kdavy_ has joined #openindiana[20:42:06] *** DontKnwMuch_ has quit IRC[20:53:37] *** sdward has joined #openindiana[20:56:33] *** Botanic has joined #openindiana[20:56:33] *** Botanic has joined #openindiana[20:57:45] *** sudosu has quit IRC[21:01:41] *** bens1 has quit IRC[21:10:07] *** sdward has left #openindiana[21:15:39] *** Naresh has joined #openindiana[21:26:21] *** spanglywires has joined #openindiana[21:32:24] *** tsukasa` has quit IRC[21:35:03] *** tsukasa has joined #openindiana[21:36:23] *** laserbled has quit IRC[21:37:42] *** lblume has quit IRC[21:41:54] *** spanglywires has quit IRC[21:48:01] *** kart_ has quit IRC[21:52:38] *** _delirium has joined #openindiana[21:54:04] <_delirium> Just install OpenIndiana, I was previously on OpenSolaris snv_134, only issue so far is that if I do an 'init 6' then the system locks up at the kernel stage during the boot. But if i power off and power back on 'init 5' then it always boots fine - should i try upgrading from oi_148 to oi_151?[22:01:03] <Alasdairrr> _delirium: its worth trying[22:01:19] <Alasdairrr> keep in mind that the /dev-il repo is a staging area for the final release and has known bugs[22:01:40] <Alasdairrr> it should work fine in most circumstances[22:04:09] <yalu> this is probably an annoying question, but is a release of 151 in sight yet?[22:04:23] <_delirium> Alasdairrr: is there a published list of know issues?[22:05:46] <Alasdairrr> _delirium: yes[22:05:51] *** ivo_ has joined #openindiana[22:05:52] <Alasdairrr> 1203,1164,1161,1134,1094,985[22:05:57] <Alasdairrr> www.illumos.org/issues[22:06:01] <Alasdairrr> yalu: We're hoping 1-2 weeks[22:06:12] <yalu> yay[22:06:13] <Alasdairrr> Future releases should be a lot quicker[22:06:22] <Alasdairrr> this one was a biggie[22:06:30] <Alasdairrr> due to the switch to Illumos and what not[22:06:38] *** raichoo has quit IRC[22:06:45] *** raichoo has joined #openindiana[22:07:26] <_delirium> Alasdairrr: okay so none of this looks too serious[22:08:07] <Alasdairrr> no, nothing too bad[22:08:11] <Alasdairrr> we've just had a major bug fixed today too[22:08:23] <Alasdairrr> 1203[22:08:31] <Alasdairrr> the description is deceptive, 1203 affects a lot of things[22:09:44] <_delirium> Alasdairrr: so if i up to 151, is 1203 fix in there?[22:10:09] <Alasdairrr> not yet[22:10:23] <Alasdairrr> it'll only bite you really if you use perl lots or foreign locales[22:12:06] <_delirium> okay i dont so no worries[22:14:02] <yalu> export LANG=C ;)[22:14:55] <yalu> I'm at the moment busy testing out how well 148 performs for deletes and rsync versus NCP and OI151 (the dev build)[22:15:19] <_delirium> Alasdairrr: here goes nothing ;-)[22:16:01] <Alasdairrr> yalu: i can't imagine there being a difference[22:18:27] <yalu> well there is. OI is a lit slower for deletes and a bit slower for rsync (5 to 10%). I'm testing with dedup on, have to admit the testing setup isn't ideal though.[22:19:37] <yalu> for deletes I measured 10 versus 17 minutes on a certain directory[22:20:10] <Alasdairrr> weird[22:21:25] <yalu> well yes. I was surprised too but apparently Nexenta has done some optimisations to this it seems. either that or I'm more lucky with my arc cache one time versus the other (I'm running this in a VM that I haven't given enough memory for the ddt).[22:22:29] <richlowe> Nexenta are running older software[22:22:36] <richlowe> then patched to become new-ish, and then changed.[22:23:14] <yalu> today to my suprise I found out that on NCP you can't do "zfs allow" because of a python bug[22:23:33] <richlowe> illumos-ishly, it's no longer python[22:23:37] <richlowe> but I'm not surprised.[22:23:46] <richlowe> I doubt they even noticed it was python until it was too late[22:23:55] <yalu> I really hope dedup is more stable on OI151 than on NCP[22:24:09] <Alasdairrr> yalu: the kernel between 148 and 151 is not massively different[22:24:12] <yalu> I once had 3 lockups in 2 days. really not funny[22:24:25] <Alasdairrr> were those lockups during zfs destroys?[22:24:40] <yalu> god forbid no. during normal deletes using oldfashioned rm[22:25:17] <Alasdairrr> i haven't heard of lockups due to dedupe during normal use[22:25:19] <Alasdairrr> strange[22:25:33] <yalu> what is anormal use? :)[22:25:45] <Alasdairrr> using it[22:25:46] <Alasdairrr> :p[22:26:02] <richlowe> Alasdairrr: NCP is 134+lots of patches[22:26:13] <richlowe> in theory, dedup should be more stable but not much faster.[22:26:14] <Alasdairrr> richlowe: yeah, I have applied those same patches to 134 myself[22:26:17] <richlowe> but then untouched since.[22:30:40] <yalu> it's a wonderful workd[22:31:08] <yalu> by the way: are deletes sync writes?[22:33:13] <tsoome> deletes do only metadata update[22:47:11] <miine> hmm. seems that package publisher are not really working/usefull...[22:47:47] <_delirium> are there any good 3rd party package repos that i should know about?[22:48:40] <miine> even when specifying a dependency with publisher (like pkg://xxx/library/openldap ) it still won't install if there is already a different library/openldap package installed...[22:49:25] <miine> _delirium: depends on how you define good... opencsw, blastwave just spring to my mind.[22:49:52] *** InTheWings has quit IRC[22:50:56] <miine> I don't know if this is in > 148 still the case because I don't know if this pkg handling is intentional or a bug... I would say its a bug.[22:51:16] <_delirium> miine: sure i know those but i meant publishers that are compatible with pkg, there used to be a home unix one that published builds of the spec files for all the ffmpeg and codec stuff, but it seems to have vanished recently[22:51:42] <richlowe> miine: an obvious argument is that packages with the same name and version should be practically _the same_[22:52:15] <_delirium> miine: spoke too soon - found it again - http://solaris.homeunix.com/?q=node/33[22:52:32] <miine> richlowe: why then introducing the publisher if it is not working ?[22:52:58] <richlowe> miine: Because you want to install software from multiple places.[22:53:00] <miine> richlowe: this means that the publisher has to use a "prefix" in his package names[22:53:09] <richlowe> and it works in every situation _other_ than the depend action (I guess, based on what you're saying)[22:53:16] <richlowe> I didn't actually know that didn't work, and am checking why.[22:53:42] <richlowe> miine: it only means that if you insist in your dependencies that you want packages from that publisher. I don't understand why you would do that[22:53:53] <richlowe> if you want foo-5.4 why does it matter where it comes from?[22:54:09] <richlowe> if it's different feature flags and such, that might actually _be_ a good reason to name it differently.[22:54:16] <richlowe> but I'm not sure, I didn't think hard about it.[22:54:30] <miine> richlowe: because I want to depend on my self build packages which might have the same name...[22:54:55] *** _delirium1 has joined #openindiana[22:55:09] <miine> richlowe: it's not about features, it's about name space. either it's working or not.[22:55:19] <miine> richlowe: and in this case the name space is not working[22:55:29] <richlowe> for dependency actions, no. as I said, I'm asking why.[22:55:58] <miine> richlowe: you ask me?[22:56:01] <richlowe> No[22:56:04] <richlowe> I'm asking the pkg(5) folks[22:56:21] <richlowe> They are also not isolated in the sense that you could install both at the same time[22:56:36] <richlowe> the publisher is "A place you get the package from", it is not part of the package name[22:56:38] <richlowe> that's the other bit.[22:56:41] *** _delirium has quit IRC[22:56:52] <richlowe> so pkg://foo/bar/baz at 1 dot 0 and pkg://blum/bar/baz at 1 dot 0 are the same package[22:56:54] <richlowe> from two sources[22:57:02] <Meths> miine: What's the reason given by pkg for not installing? Is the other publisher non-sticky? Is your package newer?[22:57:02] <richlowe> same package name and version that is.[22:57:11] *** mikw has joined #openindiana[22:57:15] <miine> pkg://de.little-iron/library/openldap at 2 dot 4.23,5.11:20110722T120847Z: This version is excluded by installed incorporation pkg://openindiana.org/consolidation/sfw/sfw-incorporation at 0 dot 5.11,5.11-0.148:20101125T013204Z[22:58:16] <richlowe> Then that is what I said above[22:58:19] <richlowe> those two packages are the same package[22:58:20] <miine> its not the same version. its just the same name: library/openldap 2.4.21-0.148 installed[23:00:22] *** _delirium1 has quit IRC[23:00:44] <miine> richlowe: why should two publisher release the same package? And if it doesn't matter (because they have to be the same) why can I specify the publisher at all within the pkg URI ?[23:00:48] <Meths> Yeah, so pkg isn't seeing yours as newer. Removing sfw-incorporation may solve the issue. Fixing your versioning would mean you won't end up removing all your consolidations as you get different overlaps.[23:02:18] <richlowe> Meths: is seeing it as newer, but prevented from installing by sfw-incorporation[23:02:20] <miine> Meths: its a pkg bug - it's not working as its meant to be. the pkg uri specifies the package. so a pkg://foo/bar/baz can hold something totally different then pkg://blum/bar/baz .[23:02:33] <richlowe> Meths: if the paths vary, the upgrade would break shit.[23:02:40] <richlowe> miine: that is not true[23:03:22] <miine> richlowe: why not? foot != blum .[23:03:46] <richlowe> because that is not part of the package name, and the pkg folks decided that be so[23:03:52] <miine> richlowe: also that means that every pkg publisher has to use it's own prefix to be sure that it is working[23:03:58] <richlowe> basically, it isn't a bug because they decided they want it to work that way.[23:04:10] *** _delirium1 has joined #openindiana[23:04:17] <miine> richlowe: is there documentation about it?[23:04:29] <richlowe> It might be in pkg(5)[23:04:39] <richlowe> but I just suggested the guy writing their manual make sure that there is documentation about it.[23:04:43] <richlowe> so "hopefully there will be"[23:05:56] <_delirium1> im i was doing a pkg image-update - via ssh and i lost the connection during the 'Update Phase' what does this mean for the state of my system?[23:06:25] <richlowe> should be fine[23:06:32] <richlowe> if it was an operation requiring a new BE, certainly is fine[23:06:37] <richlowe> though the BE it created won't be[23:06:51] <_delirium1> richlowe: okay so i can drop the be - and yes i asked for a new be[23:07:02] <richlowe> if it didn't require a new BE, it should have created a snapshot of the current BE to which you could roll back :)[23:07:32] <_delirium1> richlowe: in zfs list i just see the name that i specified to --be-name[23:07:41] <_delirium1> richlowe: so i should drop that zfs?[23:11:08] <miine> richlowe: but still I declare it a bug: it doesn't make sense this way when you see oi 148 and sol11express. both have the same package names, but different publishers... so both should rename all their packages by prefixing either oi or oracle...[23:11:53] *** DontKnwMuch has joined #openindiana[23:14:07] <Alasdairrr> _delirium1: you'd use beadm to delete the boot environment[23:14:17] <Alasdairrr> just type "beadm" and hit enter, it'll spew back the options[23:14:31] <Alasdairrr> miine: thats not a bug[23:14:52] <mnaser> how i feel after i found zfs = http://www.youtube.com/watch?v=aOl4oeHZnBk[23:14:54] <Alasdairrr> miine: They are competing operating systems that deliver similarily named packages[23:15:30] <Alasdairrr> In theory you can just depend on the package you need by name, you don't and shouldn't be more specific than that[23:15:53] <_delirium1> Alasdairrr: perfect thanks[23:15:56] <Alasdairrr> Although OI is set to diverge from Solaris 11 quite substantially in the near future[23:16:00] <Alasdairrr> we're forking[23:16:15] <Alasdairrr> in the name of progress and rapid development[23:16:23] <Alasdairrr> we can't build a community so long as we're being spoon fed[23:16:26] <alanc> it's already divergent from what will be S11 due to changes in the un-open code[23:16:36] <Alasdairrr> indeedeo[23:17:04] <miine> Alasdairrr: nope. thats not true. because the publisher IS PART of the package uri, and if I declare dependencies which include the publisher, what would you expect pkg to do?[23:17:04] <richlowe> alanc: "un-open", you could be in marketing.[23:17:06] <_delirium1> but will new techs added to Solaris Express also be added to OpenIndiana[23:17:14] <richlowe> alanc: "communally challenged"[23:17:22] <DontKnwMuch> after installing a new install, and doing a pkg image-update I am getting a lot of: build_devlink_list: readlink failed for /dev/zcons/alasdair.fastdev01.uk.openindiana.org/zoneconsole: No such file or directory - how to get rid of that[23:17:22] <alanc> yeah, I was just wondering where that came from[23:17:34] <richlowe> alanc: "cooperatively deficient"[23:17:36] <Alasdairrr> miine: The publisher isn't part of the package name really[23:17:40] <alanc> "source visibility challenged"[23:18:12] <Alasdairrr> In the URI it's a way of referring to a package from a particular publisher, it's not the definition of the package[23:18:15] <miine> so the standard packages a wasting package name space by NOT prefixing ![23:18:17] <Alasdairrr> you can move packages between repos and publishers[23:18:52] <mnaser> i was wondering how's openindiana doing with it's current mirrors[23:18:58] <mnaser> aka is it an issue?[23:19:08] <Alasdairrr> mnaser: Badly - people set them up then wandered off to resume $life-activities[23:19:13] <alanc> ooh, "source invisible code" - make it really sound like a feature, "it's not closed, just invisible to use of our amazing new cloaking device!"[23:19:34] <mnaser> Alasdairrr: is there an active "central" mirror monitor?[23:19:40] <Alasdairrr> yes i think someone set that up![23:19:59] <Alasdairrr> someone set up mirmon[23:20:05] <Alasdairrr> Is it hidden from the public?[23:20:22] *** McBofh has quit IRC[23:20:25] *** Burner_ has joined #openindiana[23:20:26] <mnaser> we use zfs for our storage and we're a pretty big provider so maybe I can figure out something to have a mirror setup[23:20:27] <Alasdairrr> this is where my shitty memory is really annoying[23:22:36] <Alasdairrr> http://91.194.75.33:8888/mirmon/[23:22:54] <Alasdairrr> 4 good, 4 bad[23:23:43] <miine> Alasdairrr: I you want to move packages between REPOS, just set them to the SAME publisher ![23:24:46] <Alasdairrr> miine: I don't understand what you're trying to achieve?[23:24:56] <Alasdairrr> The current system is by design, and works fine for most use cases[23:25:43] <miine> Alasdairrr: OK. thats really simple. I I have to prefix all my own packages and that exact prefix is taken by the oi or solaris installation later, I have to rename the prefix. that would lead to a big mess...[23:26:05] <Alasdairrr> Why do you have to prefix things?[23:26:17] <miine> Alasdairrr: I thought that I could prevent that by specifing the publisher within the dependencies...[23:26:33] <Alasdairrr> The idea of publishers is that each person that publishes software is a publisher[23:27:12] <miine> Alasdairrr: If I don't prefix, I cant be sure that my package name bar/baz won't collide with an already existing package name....[23:27:47] <Alasdairrr> That's a naming problem[23:27:50] <miine> Alasdairrr: at the current implementation it doesn't make sense at all to use a publisher within the pkg uri.[23:28:03] <Alasdairrr> not a problem with the packaging system's concept of publishers[23:28:15] <Alasdairrr> its the same issue with apt, yum, etc[23:28:16] <miine> Alasdairrr: yes. it is a naming problem which could be solved if pkg would respect the package publisher within the uri.[23:28:37] <Alasdairrr> but sometimes you want package names to be the same[23:28:52] <Alasdairrr> for example, during development, we use different publishers, like illumos-gate or pkg5-gate[23:29:02] <Alasdairrr> those publishers deliver packages with the same name[23:29:11] <Alasdairrr> that lets us upgrade easily to the latest development bits[23:29:19] <Alasdairrr> what you're proposing would shit all over that, i think[23:29:24] <miine> Alasdairrr: if you want to install a package/declare a dependency and it shouldn't matter from which publisher it comes, the publisher is simply left away...[23:29:28] <Alasdairrr> you just need to pick a suitably unique name[23:29:34] <Alasdairrr> correct[23:30:01] <Alasdairrr> depend fmri=pkg:/somepackage[23:30:09] <Alasdairrr> thats how it works - dependencies don't include the publisher[23:30:15] <miine> Alasdairrr: and who on earth says that this "unique" name isn't taken later by oi ??[23:30:38] <Alasdairrr> that's more of a social problem than a technical one[23:31:06] <miine> Alasdairrr: you won't expect to find the same content under http://foo/bar/baz and http://blum/bar/baz .[23:31:07] <tomww> hm. I thought if you insert the publisher name into the fmri then it is honoured...[23:31:10] *** hsp has quit IRC[23:31:52] <Alasdairrr> tomww: I haven't tried it, but you couldn't install both packages at the same time though and it probably isn't a good idea to have two active publishers delivering different packages with the same name[23:32:07] <Alasdairrr> if your preferred publisher has a higher version number and you image-update, for example[23:32:10] <Alasdairrr> that'd mess things up quite a bit[23:32:25] <tomww> example .p5i links[23:32:37] <Alasdairrr> if your publisher is set non-sticky of course[23:32:41] <Alasdairrr> if its sticky i guess that wouldn't happen[23:32:54] <tomww> right, only one of those packages with the same name can be installed at a time[23:33:14] <Alasdairrr> *nods*[23:33:33] <Alasdairrr> miine: thinking about it, it wouldn't be such a big issue really[23:33:54] <Alasdairrr> if you really care, your dependencies could include your publisher name[23:34:28] <Alasdairrr> if there is a name collision, you can't install both at the same time[23:34:43] <Alasdairrr> and you can just pick the one you want to install by using the publisher in the pkg uri[23:35:03] *** hsp has joined #openindiana[23:37:24] <miine> Its a bug. Take an OS who does it right: MacOS X. that would be like the preferences of //foo/bar/baz are the same as the preferences of //blum/bar/baz .[23:38:00] <miine> Of course they are not. the publisher should be used as part of the package name if given.[23:38:53] <miine> so even if library/openldap is installed, I still could install //xxx/library/openldap . Because they are different packages.[23:39:22] <Alasdairrr> then prefix[23:39:42] <miine> Alasdairrr: why doesn;t oi or oracle prefix ??[23:39:49] <Alasdairrr> because we are the distro[23:39:51] <Alasdairrr> we don't have to[23:39:54] <Alasdairrr> does apple prefix?[23:39:54] <Alasdairrr> no[23:39:57] <Alasdairrr> they use library/openldap[23:40:07] <Alasdairrr> you are mistaking what a publisher is[23:40:08] <miine> Alasdairrr: yes. APPLE does prefix everything with com.apple[23:40:21] <tomww> I think there is plenty of written stuff how and why the url to packages are designed that way.[23:40:39] <Alasdairrr> i haven't even seen an apple "package" as such, I thought you just dragged things to the applications folder[23:40:42] <Alasdairrr> but there we go[23:40:43] <Alasdairrr> i have stuff to do[23:40:46] <Alasdairrr> anyway[23:40:50] *** Alasdairrr is now known as AlasBusy[23:41:52] <miine> if there must be prefixes: why isn't there a "prefix" registration to avoid namespace collisions? same for the /opt directory name space.[23:42:06] *** hsp has quit IRC[23:43:36] <miine> and just for the record Alasdairr (if you read it later): just look into the Application.app/Contents/Info.plist (THAT is what matters in MacOS X) - there is everywhere a prefix for declaring which stuff belongs to what (like document type <-> application).[23:43:38] *** hsp has joined #openindiana[23:47:19] <miine> and by the way. I you buy a book from "random house" publisher which is called library/openldap, do you expect the same content from publisher "deutscher verlag" ?[23:48:28] <quasi> miine: maybe that's exactly what one should expect from "random house" ;)[23:49:37] <miine> quasi: the problem is that the pkg search algo is fucked up. and some here are defending it as a feature.[23:50:05] <miine> and saying that other package systems behave the same way... oh we should jump out of the window too...[23:50:48] <tsoome> why are you complaining? :P[23:51:37] <_delirium1> so I tried to upgrade from 148 to 151 - and got this - http://pastebin.com/heZUBWZm[23:51:39] *** wonslung has quit IRC[23:51:40] <_delirium1> help[23:51:45] <miine> tsoome: because I just have removed my prefix and specified the publisher name in the dependencies :-([23:52:18] <miine> tsoome: but the more you think about it, you'll see that my position is right.[23:52:30] <miine> tsoome: it's a bug in pkg.[23:52:43] *** wonslung has joined #openindiana[23:52:50] *** wonslung has quit IRC[23:52:56] <tsoome> there are many nonsenses around and not fixed:P[23:53:35] <miine> I will check now if solaris 11 express behaves the same way.[23:54:18] <miine> tsoome: but if it is nonsense, then no need to defend that[23:55:12] <miine> at the end: IF the current implementation is "right" - what would that lead to?[23:55:31] <miine> everyone has to prefix with his domain name to be sure to be unique.[23:56:09] <miine> so a pkg would be when including the publisher: pkg://com.domain/com.domain/library/openldap[23:56:15] <DontKnwMuch> hi people, remember my problems, I still have a cpu "leak", a day or so for 3% cpu usage increase, and /opt/DTT/hotkernel completely locks the sistem down (stops responding to pings..), after a fresh reboot it works ok, and even the hotkernel does not hang it, until a few days later... How can I go to find the cause?[23:56:19] <miine> looks a bit redundant...[23:56:45] <miine> DontKnwMuch: no network loop?[23:56:51] <_delirium1> anyone know what to do from here then - http://pastebin.com/heZUBWZm[23:58:48] <DontKnwMuch> network loop I did not think about... have aggregated two nics, and dladm shows http://www.pastie.org/2256425 which looks normal to me... on the switch side too.. is there some way of checking for it?[23:59:22] <miine> DontKnwMuch: didn't someone say to look at the next hop count or something like that?[23:59:58] <miine> DontKnwMuch: but I don't believe in that network loop too. that kind of packet storm should occur much faster...