September 14, 2013 
< | 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 | >
September 14, 2013 [00:00:04] <rmustacc> http://src.illumos.org/source/xref/illumos-gate/usr/src/cmd/svc/svcadm/svcadm.c[00:00:14] <rmustacc> The SMF code is not great.[00:03:32] <yo61> So _smf_repository_switch calls _scf_repository_switch[00:03:58] <yo61> from libscf.h possibly? or maybe libscf_priv.h?[00:03:59] *** jim80net has quit IRC[00:05:43] <rmustacc> That's implemented in lib/libscf/common/lowlevel.c[00:05:49] <rmustacc> Which appears to make some door call to configd.[00:06:00] *** bixu has quit IRC[00:06:04] <rmustacc> Let me save you a bit and find the right entry point in configd.[00:06:08] <yo61> k[00:08:02] <yo61> I am now officially out of my comfort zone :)[00:09:15] <yo61> What is a door call?[00:09:29] <rmustacc> Basically a means of doing a remote procedure call.[00:10:33] <yo61> Ah, OK[00:10:51] <rmustacc> In this case specifically to another process on the system.[00:12:25] <rmustacc> So I found the code that emits that error.[00:13:20] *** vizant0 has joined #smartos[00:13:47] *** julianduque has quit IRC[00:14:00] <rmustacc> http://src.illumos.org/source/xref/illumos-gate/usr/src/cmd/svc/configd/backend.c#1127[00:14:01] <yo61> Ooh, OK.[00:14:45] <yo61> So it's failing a file rename[00:14:51] <rmustacc> So now we have a bunch more questions to go ask and answer.[00:16:06] <rmustacc> So I think the first question I have is what are the permissions on /etc/svc?[00:16:49] *** andy_js has quit IRC[00:16:57] <yo61> Coming up[00:17:02] <yo61> (IN the gz or zone?)[00:17:17] <rmustacc> Zone.[00:17:25] <yo61> # ls -ld /etc/svc[00:17:25] <yo61> drwxr-xr-x 4 root sys 21 Sep 13 21:45 /etc/svc[00:17:42] <rmustacc> Okay, that looks pretty reasonable.[00:18:00] <rmustacc> Then in the zone can you run and gist something like 'ppriv $(pgrep configd)[00:18:43] <yo61> https://gist.github.com/robinbowes/6556821[00:19:30] <rmustacc> Okay.[00:20:37] <rmustacc> Well, I think it's time to do some DTrace spelunking.[00:20:49] <yo61> Oooh, I like that word[00:21:24] <yo61> Do you have children?[00:21:27] <yo61> http://vimeo.com/47648572[00:23:11] <rmustacc> So basically what we're going to do is try to figure out where in the kernel the rename is failing.[00:23:17] <yo61> OK[00:23:30] <rmustacc> I'm going to do a few iterations on a D one liner for a moment to get a sense of how many of these operations are going on.[00:23:38] <yo61> k[00:25:20] <rmustacc> So first in the global zone run this: dtrace -n 'syscall::rename:entry/execname == "svc.configd"/{ @ = count(); }' -n 'tick-5s{ printa(@); }'[00:25:26] <rmustacc> And in another shell run vmadm reboot[00:26:42] <yo61> reboot didn't work last time 0- had t stop.start - that OK?[00:26:53] <rmustacc> Sure.[00:26:58] <rmustacc> Ah yeah, that makes sense.[00:27:00] *** vizant0 has quit IRC[00:28:07] *** FatDarrel has joined #smartos[00:28:22] *** Trixboxer has quit IRC[00:28:38] <yo61> OK, so I get 3 printed every 5 secs[00:30:06] *** bixu has joined #smartos[00:30:33] <rmustacc> Okay.[00:30:37] <rmustacc> That's fine.[00:31:50] *** jelmd has quit IRC[00:32:58] <rmustacc> Okay, just need to track down a few things.[00:33:29] <rmustacc> Can you do that again with dtrace -n 'fbt::zfs_rename:entry/execname == "svc.configd"/{ @ = count(); }' -n 'tick-5s{ printa(@); }'[00:33:32] <rmustacc> Should see just three of those.[00:33:56] <yo61> OK[00:34:08] *** tbatchelli_m has joined #smartos[00:34:47] <yo61> Not seeing anything other than the tick[00:35:03] <rmustacc> So it's empty?[00:35:09] <rmustacc> Across the reboot?[00:35:14] *** Fuzai has quit IRC[00:35:15] <yo61> Looks that way[00:35:17] <rmustacc> Err, start/stop[00:36:16] <rmustacc> Can you run this: dtrace -n 'sdt:::set-error/execname == "svc.configd"/{ stack(); }'[00:36:19] <rmustacc> Across a reboot?[00:36:48] <yo61> Well, if by "empty" you meanOK[00:37:33] <yo61> Quite a lot there[00:37:43] <rmustacc> Okay, that's unfortunate.[00:37:49] <rmustacc> Let me give you a slightly different thing to run.[00:37:51] <rmustacc> One minute.[00:39:52] *** jelmd has joined #smartos[00:40:26] <rmustacc> yo61: Grab this file: wget http://us-east.manta.joyent.com/rmustacc/public/tmp/configd.d[00:40:36] <rmustacc> Run it as dtrace -s configd.d -o /var/tmp/out[00:40:49] *** mary5030 has quit IRC[00:40:54] <rmustacc> Err, actually scratch the -o /var/tmp/out from the command.[00:41:19] <yo61> OK, got one error[00:42:03] <yo61> https://gist.github.com/robinbowes/6557014[00:43:29] *** tbatchelli_m has quit IRC[00:43:35] *** Fuzai has joined #smartos[00:44:14] *** Fuzai_ has joined #smartos[00:44:15] <rmustacc> I just updated the script.[00:44:18] <rmustacc> Mind running that again?[00:44:20] <yo61> k[00:44:53] <yo61> dtrace: failed to compile script configd.d.1: line 20: undefined function name: ustacK[00:45:15] <yo61> capital 'K' ?[00:45:26] <rmustacc> Ugh, I really did that?[00:45:32] <rmustacc> Edit the script and lower case it.[00:45:43] <yo61> https://gist.github.com/robinbowes/6557051[00:48:34] <rmustacc> Okay, that one is probably actually okay.[00:48:49] <rmustacc> Let me make this script a bit more complicated.[00:50:27] <yo61> OK[00:51:32] <rmustacc> Okay, grab it again.[00:52:23] <yo61> https://gist.github.com/robinbowes/6557095[00:53:18] <rmustacc> Ugh, looks like I didn't hit save in my editor.[00:53:22] <rmustacc> Pesky :w[00:53:24] <yo61> :)[00:53:26] <rmustacc> Now try it.[00:54:51] <yo61> https://gist.github.com/robinbowes/6557115[00:57:51] <rmustacc> Hmm, okay.[00:58:02] <rmustacc> Well, looking at it that error is a red herring.[00:58:05] <rmustacc> The zfs one.[00:58:11] <rmustacc> Because we got enoent presumably trying to look that up.[00:58:14] <rmustacc> D'oh.[00:59:08] *** wolfeidau has quit IRC[00:59:31] <yo61> Back to the drawing board?[00:59:34] <rmustacc> Not quite.[00:59:41] <rmustacc> We still know that the problem is in renameat.[00:59:49] <rmustacc> But it's not getting to zfs for something.[01:00:16] *** tbatchelli_m has joined #smartos[01:03:14] *** tbatchelli_m has quit IRC[01:05:28] <rmustacc> Okay, there's a new script.[01:06:45] <yo61> https://gist.github.com/robinbowes/6557197[01:07:44] <rmustacc> Okay, well, we know what function is generating our error.[01:08:14] <rmustacc> So for some reason you're erroring in nbl_conflict.[01:08:50] <rmustacc> So I'm going to augment that script a bit more.[01:10:10] <yo61> OK[01:10:13] *** tbatchelli_m has joined #smartos[01:10:16] <yo61> (droppped off the net for a minute then)[01:11:05] <rmustacc> So I think this relates to share locking.[01:11:09] <rmustacc> And posix file locks.[01:11:21] <rmustacc> But I have no clue if anything tries to do that with the file or not.[01:11:24] <rmustacc> Which feels a bit weird.[01:11:41] <yo61> Aha[01:12:58] <rmustacc> I really have no idea about how this part of the system works. It's my first time in it.[01:13:30] <yo61> I see this in /var/adm/messages: https://gist.github.com/robinbowes/6557237[01:13:50] <yo61> Not sure if it's relevant, but you said "locking"[01:14:01] <rmustacc> It shouldn't be relevant.[01:14:05] *** FatDarrel has quit IRC[01:14:06] <rmustacc> Unless you're doing something crazy wiht nfs.[01:14:06] <rmustacc> *with[01:14:14] <yo61> Nope[01:14:14] <rmustacc> And the root of the zone.[01:14:36] <yo61> Mmm, what you you mean by that?[01:14:46] <rmustacc> If you're not using nfs it's fine.[01:14:53] <rmustacc> It'd be pretty hard to pull that off with SmartOS.[01:14:59] <yo61> I'm sharing /zones over nfs[01:15:04] <rmustacc> Oh.[01:15:08] <yo61> But not mounting anything over nfs on the smartos box[01:15:21] <rmustacc> Can you turn that off for a moment?[01:15:57] <rmustacc> Okay, now that error message is a lot more relevant.[01:16:03] <yo61> https://gist.github.com/robinbowes/6557251[01:17:13] <yo61> Turned it off[01:17:54] <yo61> THe zone start still seemed to fail[01:19:34] <yo61> Turned it back on[01:20:36] <rmustacc> I would do a full new provision with that off.[01:20:49] <rmustacc> And I'd go a step further and disable the nfs service.[01:21:11] <yo61> OK[01:21:51] <yo61> svcadm disable svc:/network/nfs/server:default ?[01:23:51] <rmustacc> Ýeah.[01:24:04] <yo61> Doesn't seem to make any difference[01:25:26] <jesse_> maybe it's the -o nbmand=on ?[01:25:37] <yo61> What's that?[01:25:47] <jesse_> it does some locking magic to sync smb and nfs shares, I think?[01:26:30] <jesse_> you pasted the zpool create[01:27:08] <yo61> But isn't all that stuff on /etc/svc/ loaded from the usbkey ?[01:27:16] <rmustacc> For the global zone only.[01:27:46] <rmustacc> /etc for a zone is on the zpool.[01:27:52] <yo61> k[01:27:53] <jesse_> and that's a zfs attribute, it might do funny things[01:28:44] <jesse_> http://www.mail-archive.com/zfs-discuss at opensolaris dot org/msg28412.html[01:29:01] <yo61> Bingo - turning off nmband seems to have worked[01:29:11] <yo61> Successfully created VM ad90152f-d1df-428e-89b0-d58cb0cdab7d[01:29:36] <wesolows> wow. I thought nbmand was only relevant for CIFS[01:29:51] <yo61> All services up and running[01:29:54] <rmustacc> So now maybe that's happening because lockd is down?[01:30:43] <yo61> What service runs lockd?[01:31:15] <yo61> OK - nlockmgr[01:31:34] <rmustacc> So here's the next test:[01:31:42] <rmustacc> Enable svc:/network/nfs/status:default[01:31:47] <rmustacc> Then svc:/network/nfs/nlockmgr:default[01:31:51] <rmustacc> Then the nfs server.[01:31:53] <rmustacc> And do a zone create.[01:32:14] <yo61> The first one is enabled[01:32:22] <yo61> (already)[01:33:11] <yo61> And the lockmgr is also enabled[01:34:24] <yo61> Still working[01:34:34] <yo61> ie. vmadm create works[01:34:49] <jesse_> nbmand enabled, too?[01:35:08] <yo61> So, that's with ... oh[01:35:12] <yo61> It was off[01:35:18] *** tbatchelli_m has quit IRC[01:35:32] <yo61> But it's now showing as zones nbmand on temporary[01:37:22] *** szaydel has quit IRC[01:37:42] <rmustacc> nmband is something I know little about.[01:38:39] <yo61> OK, so I ran: zfs set nbmand=on zones[01:38:51] <yo61> The zone create failed again[01:39:42] <rmustacc> Is there anything about the kernel lock manager in there?[01:40:01] <rmustacc> In /var/adm/messages[01:41:03] <yo61> I don't see anything[01:42:15] <rmustacc> And the lock manager is running this time?[01:42:44] <yo61> online 15:50:22 svc:/network/nfs/nlockmgr:default[01:43:16] <yo61> I re-enabled nfs sharing and the zone creation still works[01:44:23] <yo61> So, I think we have a re-producible bug/error[01:45:26] <rmustacc> I really would not recommend sharing all of /zones via nfs.[01:46:07] <rmustacc> Oh, you said it still works.[01:46:17] <rmustacc> So this all makes sense now.[01:46:21] <yo61> Yes, still works[01:46:26] <rmustacc> Lock manager was down.[01:46:31] <rmustacc> So you couldn't establish mandatory locks.[01:46:43] <yo61> Lock manager was down at boot up, but seemed to come up[01:46:56] <rmustacc> Well, the lock manager failed.[01:47:41] <yo61> Am rebooting the gz to see what comes up[01:48:29] <rmustacc> We don't enable that by default. Are you via /opt/custom/smf or something?[01:48:46] <yo61> nbmand ?[01:48:56] <rmustacc> No, the nfs server and its ilk.[01:49:03] <rmustacc> Like lockd and statd.[01:49:15] <yo61> It was already enabled in the zpool[01:49:23] <yo61> I imported an existing pool[01:49:42] <rmustacc> Those services don't come from the pool.[01:49:42] <yo61> Either that, or I manually enabled it[01:49:52] <rmustacc> Perhaps it saw the sharenfs.[01:49:55] <rmustacc> And enabled the nfs server.[01:49:59] <yo61> Well, yes, that's what I mean[01:50:17] <rmustacc> So could be a bug there then. That it's not properly enabling statd + lockd.[01:50:22] <yo61> That is one of the things I really like about using ZFS for NFS[01:50:27] <rmustacc> As that was the failure.[01:50:38] <yo61> set sharenfs=on => nfs server is enabled[01:51:26] <rmustacc> But what about the statd and lockd services?[01:52:00] <yo61> Perhaps enabled by default?[01:52:09] <yo61> Sorry, let me say what I really mean[01:52:12] <rmustacc> They aren't.[01:52:31] <yo61> Perhaps enabled along with the nfs server, etc. when sharenfs is detected ?[01:52:32] <rmustacc> When you booted the system previously we saw that lockd was running and had errored out, but statd was never started.[01:53:01] <rmustacc> Because lockd couldn't talk to stat it exitedd.[01:53:24] <yo61> That seemed to clear, ie. those services started[01:53:26] <rmustacc> Therefore all the nbmand exited.[01:53:38] <rmustacc> On the next boot?[01:53:51] <yo61> No, some time later that boot[01:53:57] <yo61> Er, that run[01:54:45] <yo61> I see the same errors after this reboot:[01:54:45] <yo61> 2013-09-13T23:50:32.830059+00:00 nas02 klmmod: [ID 814159 kern.notice] NOTICE: Failed to connect to local statd (rpcerr=5)#012[01:54:49] <yo61> 2013-09-13T23:50:32+00:00 nas02 /usr/lib/nfs/lockd[2461]: [ID 491006 daemon.error] Cannot establish NLM service over <file desc. 9, protocol udp> : I/O error. Exiting[01:54:52] <yo61> 2013-09-13T23:50:32+00:00 nas02 svc.startd[8]: [ID 652011 daemon.warning] svc:/network/nfs/nlockmgr:default: Method "/lib/svc/method/nlockmgr" failed with exit status 1.[01:55:28] <yo61> However, both statd and nlockmgr are now online[01:56:00] <rmustacc> Probably a race condition in boot.[01:56:20] <wesolows> new lock manager, new bugs[01:56:46] <rmustacc> Either that or just service race condition.[01:56:56] <wesolows> yeah, could have been there for 10 years[01:57:06] <yo61> So, I can now create zones[01:57:16] <rmustacc> Glad we were able to get to the bottom of that.[01:57:24] <yo61> Thanks ever so much for the help[01:57:28] <wesolows> I've been watching and worrying over this all day :)[01:57:43] <yo61> I've been stumped by it for months![01:57:47] <rmustacc> Well, ultimately I'm glad this is the bug.[01:57:57] <rmustacc> Oh this has been happening for a while?[01:58:02] <yo61> Yeah[01:58:16] <wesolows> never mind then. :-/[01:58:17] <yo61> I've not been able to create zones for month[01:58:19] <yo61> s[01:58:32] <rmustacc> Well, that makes it more likely to be the service race conditions vs. the new klm[01:58:47] <rmustacc> So we're still not 100% at root cause.[01:58:49] <yo61> Obviously, it's not been all that critical[01:59:10] <rmustacc> What we need to do is figure out how the services allowed this to happen.[01:59:13] <yo61> ie. I've just managed without native zones[01:59:53] <yo61> I think the service thing is different to the nbmand=on error[02:00:29] <rmustacc> Well, the main difference we've detected is that when the lock manager is off and nmband is on you can't provision zones.[02:00:39] <yo61> If I do: "zfs set nmband=on zones", it breaks zone creation[02:00:55] <yo61> No, the lock manager is on[02:01:00] <rmustacc> Huh, oh.[02:01:21] <yo61> There *is* some race condition at boot, but it sorts itself out[02:01:22] <rmustacc> I swore you had just said zone creation worked with that on a moment ago.[02:02:01] <yo61> What happens is that if you turn it off explicitly, it is still on, but listed as being "temporary"[02:02:37] <yo61> Or at least, that was what I saw before the reboot[02:03:18] <yo61> It's now showing:[02:03:19] <yo61> NAME PROPERTY VALUE SOURCE[02:03:19] <yo61> zones nbmand off local[02:03:23] <yo61> It was showing:[02:03:24] <yo61> NAME PROPERTY VALUE SOURCE[02:03:24] <yo61> zones nbmand off local[02:03:28] <yo61> Oops[02:03:33] <yo61> NAME PROPERTY VALUE SOURCE[02:03:37] <yo61> zones nbmand off temporary[02:03:40] <yo61> Gah[02:03:48] <yo61> NAME PROPERTY VALUE SOURCE[02:03:57] <yo61> zones nbmand on temporary[02:04:22] <yo61> It's late![02:05:30] <rmustacc> wesolows: Maybe you understand a bit more about how nwband works?[02:07:29] <wesolows> not really. it's "non-blocking mandatory locking"[02:07:53] <wesolows> I thought it was a feature they added to support CIFS (once and now SMB) only[02:10:58] <yo61> I've updated https://github.com/joyent/smartos-live/issues/261[02:11:48] <yo61> And I think I'm going to go to bed![02:15:41] <rmustacc> Take care.[02:15:49] <yo61> Thanks again for your help[02:17:33] <rmustacc> Sure, no problem.[02:23:18] *** andoriyu has quit IRC[02:27:31] *** szaydel has joined #smartos[02:33:42] *** elijah-mbp has quit IRC[02:40:03] *** mary5030 has joined #smartos[02:52:53] *** potatosalad has quit IRC[02:54:10] *** FatDarrel has joined #smartos[03:01:57] *** Fuzai has quit IRC[03:01:57] *** Fuzai_ has quit IRC[03:07:20] *** wolstena has quit IRC[03:15:38] *** Fuzai has joined #smartos[03:16:10] *** Fuzai_ has joined #smartos[03:17:11] *** FatDarrel has quit IRC[03:19:47] *** utlemming has quit IRC[03:20:05] *** bixu has quit IRC[03:27:58] *** szaydel has quit IRC[03:28:36] *** szaydel has joined #smartos[03:46:29] *** fxhp has quit IRC[03:48:37] *** fxhp has joined #smartos[04:01:11] *** ilovezfs has joined #smartos[04:06:34] *** Fuzai has quit IRC[04:06:34] *** Fuzai_ has quit IRC[04:08:38] *** Fuzai has joined #smartos[04:09:14] *** Fuzai_ has joined #smartos[04:31:57] *** ira has quit IRC[04:43:25] *** joffe has quit IRC[04:44:21] *** Fuzai_ has quit IRC[04:44:21] *** Fuzai has quit IRC[04:44:47] *** Fuzai has joined #smartos[04:44:49] *** Fuzai_ has joined #smartos[04:58:14] *** mattconn_ has joined #smartos[05:01:26] *** mattconnolly has quit IRC[05:18:13] *** mary5030 has quit IRC[05:33:34] *** andoriyu has joined #smartos[06:19:57] *** jaimef has quit IRC[06:23:31] *** jaimef has joined #smartos[06:31:52] *** mattconn_ has quit IRC[06:42:55] *** nikolam_ has joined #smartos[06:43:12] *** mattconnolly has joined #smartos[06:43:57] *** nikolam has quit IRC[07:05:37] *** mattconnolly has quit IRC[07:24:58] *** nefilim has quit IRC[07:39:29] *** tonist has joined #smartos[07:53:19] *** Alek2 has joined #smartos[07:53:21] *** hotbox_ has joined #smartos[07:54:05] *** griffin_ has joined #smartos[07:54:42] *** Alek has quit IRC[07:55:09] *** hotbox has quit IRC[07:58:00] <griffin_> so this is a real shot in the dark here, I have a ASMedia 1061/2 pci-e esata card, connected to a sil3726 port multiplier, typically in the past to get this setup working I have disabled NCQ on asm1061 (freebsd via adding a noncq quirk to the source), is this possible somehow on smartos?[07:59:37] *** szaydel has quit IRC[08:09:25] *** mamash has joined #smartos[08:09:52] <griffin_> zfs_vdev_sync_read_max_active/w0t1 zfs_vdev_async_read_max_active/w0t1 zfs_prefetch_disable/W0t0 , gets me from 4M/sec reads to 75M/sec reads, but it seems like overkill globally getting rid of ncq[08:41:11] *** nikolam_ has quit IRC[09:20:02] *** texarcana has quit IRC[09:22:07] *** texarcana has joined #smartos[09:24:26] *** andoriyu has quit IRC[09:24:47] *** mattconnolly has joined #smartos[09:29:02] *** mattconnolly has quit IRC[09:48:56] *** nikolam has joined #smartos[09:59:33] *** chi has quit IRC[09:59:34] *** chi has joined #smartos[10:03:52] *** jaimef has quit IRC[10:08:33] *** jaimef_ has joined #smartos[10:13:44] *** andy_js has joined #smartos[10:38:07] *** xinkeT has joined #smartos[10:41:02] *** Tekni has quit IRC[10:58:57] *** sparq_ has joined #smartos[11:12:57] *** mattconnolly has joined #smartos[11:17:38] *** mattconnolly has quit IRC[11:30:09] *** mattconnolly has joined #smartos[11:38:17] *** mamash has left #smartos[11:44:38] *** wiedi has joined #smartos[12:06:38] *** sparq_ has quit IRC[12:15:30] *** trentster has quit IRC[12:20:10] *** trentster has joined #smartos[12:32:19] *** trentster has quit IRC[12:37:05] *** indutny has joined #smartos[12:37:13] *** trentster has joined #smartos[12:38:39] *** nikolam has quit IRC[12:49:38] *** nikolam has joined #smartos[12:51:43] *** ilovezfs has quit IRC[12:53:54] *** trentster has quit IRC[12:55:23] *** ilovezfs has joined #smartos[12:56:27] *** estibi has joined #smartos[12:57:36] *** ilovezfs_ has joined #smartos[12:59:17] *** trentster has joined #smartos[13:00:12] *** ilovezfs has quit IRC[13:00:12] *** ilovezfs_ is now known as ilovezfs[13:10:25] *** jellydonut has joined #smartos[13:18:08] *** ilovezfs has quit IRC[13:21:45] *** ilovezfs has joined #smartos[13:26:50] *** Fuzai has quit IRC[13:35:49] *** mgls has joined #smartos[14:56:48] *** szaydel has joined #smartos[15:11:41] *** PurpleHaze has joined #smartos[15:12:08] *** PurpleHaze has left #smartos[15:22:17] *** mary5030 has joined #smartos[15:25:51] *** szaydel has quit IRC[15:31:02] *** mary5030 has quit IRC[15:31:14] *** szaydel has joined #smartos[15:31:28] *** mary5030 has joined #smartos[15:38:33] *** estibi has quit IRC[15:42:16] *** mary5030 has quit IRC[15:48:01] *** tonist has quit IRC[15:51:44] *** tonist has joined #smartos[16:08:52] *** andy_js has quit IRC[16:16:11] *** andy_js has joined #smartos[16:31:15] *** nefilim has joined #smartos[16:40:39] *** szaydel has quit IRC[16:53:03] *** djun-bb has joined #smartos[16:55:52] *** tonist has quit IRC[16:56:23] *** djun-bb2 has joined #smartos[16:59:43] *** djun-bb has quit IRC[17:02:57] *** szaydel has joined #smartos[17:09:34] *** nefilim has quit IRC[17:11:30] *** szaydel has quit IRC[17:14:54] *** tonist has joined #smartos[17:15:08] *** potatosalad has joined #smartos[17:32:41] *** tonist has quit IRC[18:07:52] *** potatosalad has quit IRC[18:16:27] *** vizant0 has joined #smartos[18:25:22] *** potatosalad has joined #smartos[18:37:30] *** leecallen has quit IRC[18:37:44] *** leecallen has joined #smartos[18:38:09] *** nikolam has quit IRC[18:47:54] *** nikolam has joined #smartos[18:53:11] <jesse_> jperkin, man gcc seems to be borken on multiarch[18:54:54] <jesse_> jperkin, uhm, make that broken on multiarch zsh, bash gives error and manpage[18:55:19] <jesse_> and tells me about pkg_alternatives... that should be used for java, too?[19:04:15] <jesse_> isn't gcc -m64 -L/lib supposed to look for the libraries in /lib/amd64, not /lib?[19:04:37] <jesse_> (where /lib is /opt/local/lib in this case)[19:05:28] <rmustacc> jesse_: Not if you specify it yourself I think.[19:05:37] <rmustacc> But I'm not an expert on that.[19:05:46] <jesse_> so all -L lines need to be amd64fied?[19:05:57] <rmustacc> That's pretty much always been the case I thought.[19:05:58] <jesse_> that can't be right[19:06:02] <rmustacc> But wesolows would know better.[19:06:04] <Meths> (and -R)[19:06:15] <jperkin> no that's not right[19:06:39] <jperkin> if you specify -L/usr/lib yourself, then of course you will get that, but if you do nothing it will DTRT[19:07:21] <jesse_> so, if you have non-default lib directory somewhere, it will require amd64fication[19:08:06] <jperkin> well yeh, if you want to use both ABIs[19:11:23] <jesse_> I wonder how this even compiled before on gcc 4.7...[19:13:05] *** jim80net has joined #smartos[19:20:58] *** potatosalad has quit IRC[19:35:46] *** chakatz has joined #smartos[19:36:09] <koaps> hi guys, I just made a centos KVM VM and it's taking a while login, I'm seeing this in the messages, "/lib/smartdc/mdata-get - Error getting metadata for key enable_motd_sys_info:" and "/lib/smartdc/mdata-get - Error getting metadata for key enable_motd_joyent_info: " does anyone know what the issue might be?[19:38:37] <rmustacc> koaps: If you take a look in the global zone is the metadata service in maintenance perhaps? You'd want to run svcs -xv to determine that.[19:38:52] *** prasm_ has quit IRC[19:39:32] <koaps> rmustacc: seems ok, online 16:30:24 svc:/system/smartdc/metadata:default[19:39:35] *** jedahan has joined #smartos[19:44:47] *** tonist has joined #smartos[19:45:57] <koaps> alright, I just turned off the dynamic motd and that makes login super fast now[19:51:06] <jesse_> jperkin, the reason I needed to add the -L/opt/local/lib was because without it gcc couldn't find z or iconv at all[19:51:25] <jesse_> maybe that's the real problem here?[19:51:56] <rmustacc> jesse_: That should be automatic in newer datasets.[19:52:10] *** mamash has joined #smartos[19:52:40] <jesse_> java-compile-multiarch-13.2.2 and gcc48[19:52:56] <jesse_> s/java-compile-//[19:53:25] <rmustacc> Weren't we unsure if gcc48 had that forward ported?[19:53:44] <jesse_> wasn't that something else?[19:54:06] <rmustacc> No, that was exactly the automatic adding of the -I/-L/-R flags.[19:54:50] <rmustacc> Which would do the right thing for -m32/-m64[19:56:04] <jesse_> yeah, I had to check backlog to realize it was exactly this[19:56:47] <jesse_> I was just thinking of that being adding /opt/local/lib[19:57:31] <jesse_> and -m64 +/amd64 being a separate, standard, thing[20:01:03] *** pringlescan has joined #smartos[20:01:21] <pringlescan> nahmu: pm me if you want bitbucket access, I wanted to polish a few things[20:01:26] <pringlescan> (to the atlassian installer)[20:15:13] *** potatosalad has joined #smartos[20:16:13] <jesse_> OpenJDK Server VM (24.0-b56) for solaris-x86 JRE (1.7.0_40-b43), built on Sep 14 2013 15:49:38 by "jdk" with gcc 4.8.1[20:16:36] *** ktk is now known as ktkNA[20:17:21] *** sa5bke has joined #smartos[20:18:00] <rmustacc> jesse_: Neat.[20:20:50] <jesse_> and[20:20:51] <jesse_> OpenJDK 64-Bit Server VM (24.0-b56) for solaris-amd64 JRE (1.7.0_40-b43), built on Sep 14 2013 17:30:25 by "jdk" with gcc 4.8.1[20:20:59] *** benbange1t is now known as benbangert[20:22:00] <jesse_> I still need to do a diff between pkgsrc and mine and see what jperkin has changed and how to make further updates as easy as possible[20:23:03] *** zr0 has quit IRC[20:36:56] *** prasm has joined #smartos[20:37:13] *** zr0 has joined #smartos[20:43:26] *** zr0 has quit IRC[20:46:50] *** nikolam has quit IRC[20:47:12] *** chakatz has quit IRC[20:50:07] *** leecallen has quit IRC[20:50:38] *** chakatz has joined #smartos[20:55:51] *** papertigers has joined #smartos[21:00:35] *** FatDarrel has joined #smartos[21:13:59] *** zr0 has joined #smartos[21:16:01] *** mikl has joined #smartos[21:16:07] *** trentster has quit IRC[21:17:16] *** chakatz has quit IRC[21:20:42] *** mamash has left #smartos[21:20:46] *** trentster has joined #smartos[21:21:53] *** mary5030 has joined #smartos[21:23:28] *** djun-bb2 has quit IRC[21:23:55] *** zr0 has quit IRC[21:28:24] *** zr0 has joined #smartos[21:30:14] *** djun-bb has joined #smartos[21:32:08] *** ipalreadytaken has joined #smartos[21:39:36] *** mary5030 has quit IRC[21:45:36] *** kimc has joined #smartos[21:45:37] *** kimc has joined #smartos[21:45:48] *** nikolam has joined #smartos[21:49:41] *** zr0 has quit IRC[21:52:19] *** vizant0 has quit IRC[21:52:42] *** vizant0 has joined #smartos[22:03:56] *** zr0 has joined #smartos[22:04:25] *** nathan7 is now known as Pinkie-Pie[22:04:34] *** Pinkie-Pie is now known as nathan7[22:08:36] *** tonist has quit IRC[22:09:37] *** andy_js has quit IRC[22:09:53] *** noahmehl has joined #smartos[22:21:56] *** sparq_ has joined #smartos[22:28:12] *** noahmehl_ has joined #smartos[22:28:17] *** szaydel has joined #smartos[22:28:26] *** noahmehl has quit IRC[22:28:27] *** noahmehl_ is now known as noahmehl[22:29:01] <yo61> jperkin: when's Q3 due to be released?[22:29:55] <yo61> Now that I can create zones (yay) I'm going to deploy some stuff on my smartOS box (ElasticSearch/redis/logstash)[22:30:40] <yo61> I'm wondering whetther to wait a while until you publish Q3, or to go ahead and either use Q2 or compile my own[22:31:11] *** yruss972 has joined #smartos[22:31:28] <jperkin> freeze starts tomorrow and will last for two weeks[22:31:41] *** sa5bke has quit IRC[22:31:47] <yo61> Can I see what's in the release anywhere?[22:31:49] <jperkin> after that give it a week (but hopefully less) for us to get packages ready[22:32:12] <jperkin> I don't anticipate the length of delay we had with Q2, that had a lot of changes that we made to the way the builds were done[22:32:38] <jperkin> well, with 12,000 regularly updated packages it's always going to be hard to get a concise list, and currently there isn't really one.[22:32:50] <jperkin> there is doc/CHANGES-2013 which lists every update, but that's just a firehose[22:33:03] <yo61> Ah, OK. I was wondering is there was a public repo, or something[22:33:08] <jperkin> so I tend to just try and summarise what I think are the larger changes[22:37:50] <jperkin> public repo of what?[22:38:00] <yo61> The package sources, etc.[22:38:31] <jperkin> sure, that's how you build stuff ;)[22:39:31] <yo61> Any where might I find said repo?[22:40:35] <jperkin> primary is anoncvs at anoncvs dot netbsd.org:/cvsroot -P pkgsrc, we develop with git though based on joerg@netbsd's cvs->git repo which you can find here https://github.com/jsonn/pkgsrc - ours is https://github.com/joyent/pkgsrc[22:40:56] <yo61> thx[22:41:18] <jperkin> so you could e.g. check https://github.com/jsonn/pkgsrc/commits/trunk for what's coming through - we don't always keep completely up-to-date with trunk[22:43:08] <jperkin> if you want to start building stuff from pkgsrc then my guide on doing that is: http://www.perkin.org.uk/posts/pkgsrc-on-smartos-zone-creation-and-basic-builds.html[22:43:35] *** potatosalad has quit IRC[22:44:03] <yo61> Thanks[22:44:09] <jperkin> I'd recommend, even if the stuff you want isn't currently in there, working on building stuff using pkgsrc, as it does a huge amount of work for you (fetching, patching, compiler args, etc)[22:44:54] <yo61> I want/need: elasticsearch, redis, logstash[22:46:20] <jperkin> we have redis-2.6.16, the other two you would need to write yourself[22:46:28] <yo61> OK[22:46:39] <jperkin> but we can help with that, and then once it's done can import it into pkgsrc[22:46:48] <jperkin> so everyone else gets binaries too[22:46:50] <yo61> logstash is just a single .jar, so should be pretty easy[22:47:52] <yo61> OK, thanks[22:47:54] *** potatosalad has joined #smartos[22:55:17] *** vizant0 has quit IRC[23:02:16] *** mattconnolly has quit IRC[23:03:30] *** chakatz has joined #smartos[23:04:05] *** andy_js has joined #smartos[23:10:01] *** mgls has quit IRC[23:14:41] *** ipalreadytaken has quit IRC[23:16:24] *** FatDarrel has quit IRC[23:28:06] *** szaydel has quit IRC[23:38:13] *** jim80net has quit IRC[23:39:56] *** jim80net has joined #smartos[23:42:37] *** vizant0 has joined #smartos[23:44:20] *** szaydel has joined #smartos