December 8, 2012 
< | 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 | >
December 8, 2012 [00:00:09] *** deirdres has quit IRC[00:00:10] *** deirdres_ is now known as deirdres[00:02:09] <Aram> hmm, I did some more debugging and I could step through consconfig_dacf`* until I got back to main and then the machine shut down completely (not only CPU halt).[00:05:13] <Aram> btw, I entered this case: http://src.illumos.org/source/xref/illumos-gate/usr/src/uts/common/io/consconfig_dacf.c#1370 is that normal, I don't have a serial keyboard, but perhaps that means something else.[00:05:29] <Aram> or maybe the debugger itself implements a keyboard.[00:05:55] *** jim80net has quit IRC[00:07:53] <chriss-> it decided its prompt is now a different bcom mac addr, still no active nics after boot[00:07:56] <chriss-> good times[00:07:58] <chriss-> time to go home[00:18:59] *** mamash has joined #smartos[00:21:04] *** tonyarkles has joined #smartos[00:25:02] <Aram> rmustacc: I'll try an older smartos from may, as I understand it, that's before the new acpi stuff in smartos that hasn't been integrated in illumos yet.[00:32:39] *** mamash has left #smartos[00:33:04] *** daleg has quit IRC[00:39:08] *** tonyarkles has quit IRC[00:47:08] <richlowe> Aram: did you step step step step step <gone>, or did you continue at some point?[00:47:11] <richlowe> or step over?[00:47:21] <Aram> richlowe: only step[00:47:34] <richlowe> promising, as far as isolation[00:49:50] <Aram> hmm, smartos from april hangs in the same place[00:55:19] * jeffpc is having some perf issues / stability issues on a 2 socket AMD box :([01:07:45] <konobi> Aram: did you try modifying the boot_archive to include the mouse8042 driver?[01:08:04] <Aram> sorry, not yet.[01:08:13] <Aram> tomorrow, maybe, it's 1:08AM here.[01:08:35] <konobi> heh, fair enough[01:09:55] *** deirdres has quit IRC[01:10:10] *** deirdres has joined #smartos[01:10:11] <rmustacc> jeffpc: What's your trouble?[01:10:26] <jeffpc> rmustacc: sometimes, the system locks up during boot[01:10:28] <konobi> oh... i have a machine at home that may exhibit this[01:10:33] <jeffpc> (after the SunOS banner)[01:10:52] <jeffpc> at other times, it boots, but vmadm create/delete take several minutes to complete[01:11:21] <rmustacc> Well, if locks up, nmi it.[01:11:36] <rmustacc> As for the vmadm create/delete, you'll have to see where the time is being spent.[01:11:48] <jeffpc> I did run it through truss[01:11:54] <rmustacc> I would probably pstack it or something like that.[01:11:56] <jeffpc> it seems to be stuck no a devfsadm door call[01:12:06] <rmustacc> On devfsadm/[01:12:08] <rmustacc> Hmm.[01:12:24] <jeffpc> well, devfsadmd or whatever the service is[01:13:34] *** rancor has quit IRC[01:13:57] *** rancor has joined #smartos[01:15:48] <jeffpc> rmustacc: ptree shows: http://pastie.org/5496500[01:16:18] <jeffpc> pid 3237 is stuck on /etc/dev/.devfsadm_synch_door[01:16:32] <jeffpc> so, to be correct, it's zoneadmd that's stuck[01:16:55] <rmustacc> So if it's stuck on the door, pstack away.[01:17:20] *** dysinger has quit IRC[01:19:40] <jeffpc> http://pastie.org/5496521[01:19:51] <jeffpc> (had to do it again since the first one timed out)[01:20:24] <rmustacc> the zoneadmd isn't the interesting one, what it's calling is the interesting one.[01:21:07] <jeffpc> the zoneadm (not zoneadmd)?[01:22:50] <rmustacc> No, devfsadmd.[01:22:54] <rmustacc> The place it's making the door call.[01:25:07] <jeffpc> http://old.pastie.org/5496541[01:38:22] *** rcamera has quit IRC[01:41:53] <jeffpc> http://old.pastie.org/pastes/5496608/text[01:42:09] <jeffpc> looks like a thread sits on a semaphore for way too long[01:55:23] *** Aram has quit IRC[01:56:21] <jeffpc> hrm[01:56:26] <jeffpc> ioctl that took 55 seconds[01:59:02] <rmustacc> That's suspicious.[01:59:04] <rmustacc> What icotl?[01:59:18] <jeffpc> yep, working on it[01:59:41] <richlowe> ![02:03:20] *** ahaydock has quit IRC[02:04:11] <jeffpc> http://pastie.org/5496669[02:06:58] <jeffpc> hrm...[02:07:08] <jeffpc> DIIOC | DINFOLODRV[02:07:32] <jeffpc> on /devices/pseudo/devinfo@0:devinfo, AFAICT[02:08:12] <jeffpc> http://src.illumos.org/source/xref/illumos-gate/usr/src/lib/libdevinfo/devinfo.c#108[02:08:51] <richlowe> jeffpc: fbt entry/return below devi_tree_walk , with timestamps, maybe?[02:09:05] <richlowe> (flamegraph!)[02:10:04] <jeffpc> richlowe: devi_tree_walk is in userspace...[02:10:20] <jeffpc> I suppose that's still doable[02:12:37] <richlowe> Right, I know it is, but I also don't see anything under the ioctl you mention that immediately stands out as being potentially slow[02:12:55] <richlowe> flamegraph should throw 55 seconds(!) of suck really clearly though, compared to testing guesses one by one.[02:13:33] <richlowe> much easier (and broader) to go "fbt:::entry,fbt:::return / self->trace /" with trace set under devi_tree_walk, than to screw around matching the ioctl[02:13:34] <jeffpc> I've never made one...[02:13:37] <jeffpc> yeah[02:13:43] <jeffpc> that's what I'm doing now[02:13:53] <richlowe> and I've not used the flamegraph stuff either (sorry brendan!)[02:13:58] <richlowe> but I recall it being easy and well described?[02:14:08] <konobi> canon... sun... !!![02:14:21] <jeffpc> in principle, yes[02:15:38] <jeffpc> ok...[02:16:01] <jeffpc> maybe tracing every function wasn't the best idea[02:16:10] <jeffpc> turns out lots of things happen[02:16:11] <richlowe> underneath there? It's not a bad idea[02:16:20] <richlowe> ideally, brendans magic should then bring clarity[02:17:04] <richlowe> ... presuming it can process input other than ustack() etc.[02:17:12] <richlowe> otherwise you'll have to reword the script a bit to pacify it[02:17:38] <richlowe> profile-997 /self->trace/ { stack(); } should still be valid over that much random dragging, I think?[02:20:13] <rmustacc> Yeah. It should.[02:20:21] <rmustacc> Or you could use /pid or tid/[02:20:56] <richlowe> brendang: it'd be nice to be able to format entry/return probes + timestamps into flamegraphs, though.[02:20:59] <richlowe> unless my brain has failed[02:26:58] <konobi> huh... gmake -j128 ?[02:27:29] <richlowe> konobi: someone was optimistic[02:27:58] <richlowe> that said, if your build machine isn't FSS/cpu-capped into sanity, you've lost anyway[02:29:48] <konobi> richlowe: this is in a smartos zone building smartos[02:30:06] <konobi> specifically the "bootstrap" target[02:31:11] <jeffpc> dtrace -x stackframes=100 -n 'profile-997 /pid==62/ { @[stack()] = count(); } tick-200s { exit(0); }' -o out.kern_stacks[02:31:26] <jeffpc> that produced only 26 stacks[02:32:18] <jeffpc> http://pastie.org/5496744[02:39:18] <jeffpc> hrm[02:39:24] <rmustacc> konobi, richlowe: That's in the illumos-extra Makefifile[02:39:28] <jeffpc> I just tried to make a flame graph[02:39:37] <konobi> src/Maefile.defs, yeah[02:45:59] <konobi> deirdres: it's hard to come up with a decent unix tat i think[02:47:04] <deirdres> dunno, has it been tried?[02:47:19] <richlowe> I hope not[02:47:23] <richlowe> but probably[02:47:41] <konobi> you've seen e^ipi's leg tat?[02:47:49] <richlowe> Though, if you wanted to commemerate bad decisions that last 40 years, really, is there a _better_ way?[02:48:09] <konobi> richlowe: could be worse... could be a "Zune" tat[02:52:32] *** tonyarkles has joined #smartos[02:58:50] <jeffpc> bah[02:58:58] <jeffpc> I don't see anything outright bad[03:02:46] <jeffpc> http://31bits.net/download/per-cpu-profile-997.tar.bz2[03:06:48] <konobi> richlowe: right now i'm trying to think of a good theme for a cover-up[03:15:18] *** dysinger has joined #smartos[03:25:23] *** tonyarkles has quit IRC[03:29:18] *** deirdres has quit IRC[03:57:08] *** rc10 has joined #smartos[04:03:47] *** dysinger has quit IRC[04:04:10] *** dysinger has joined #smartos[04:21:34] *** chriss- has quit IRC[04:26:05] *** chriss- has joined #smartos[04:26:36] *** dysinger has quit IRC[04:26:50] *** notmatt has joined #smartos[04:33:46] *** dap has quit IRC[05:18:18] *** notmatt has quit IRC[05:28:07] *** jgrafton has joined #smartos[05:46:33] *** notmatt has joined #smartos[05:53:43] *** rancor has quit IRC[05:58:12] *** notmatt has quit IRC[06:46:14] *** rc10 has quit IRC[06:58:37] *** rc10 has joined #smartos[07:05:34] *** rc10 has quit IRC[07:26:29] *** jamesd has quit IRC[08:20:17] *** e^ipi_laptop has quit IRC[08:21:19] *** e^ipi_laptop has joined #smartos[08:31:05] *** e^ipi_laptop has quit IRC[08:31:41] *** e^ipi_laptop has joined #smartos[08:53:05] *** fsteinelX has joined #smartos[08:57:05] *** e^ipi_laptop has quit IRC[09:01:16] <MerlinDMC> morning[09:03:48] *** e^ipi_laptop has joined #smartos[09:15:51] *** trentster has quit IRC[09:16:05] *** trentster has joined #smartos[09:17:01] *** e^ipi_laptop has quit IRC[09:17:36] *** e^ipi_laptop has joined #smartos[09:22:02] *** trentster1 has joined #smartos[09:22:50] *** texarcana has quit IRC[09:23:00] *** trentster has quit IRC[09:23:37] *** e^ipi_laptop has quit IRC[09:24:05] *** texarcana has joined #smartos[09:32:32] *** trentster has joined #smartos[09:32:59] *** trentster1 has quit IRC[09:38:34] *** trentster has quit IRC[10:00:08] *** fsteinelX has quit IRC[10:14:58] *** cncfanatics has joined #smartos[10:50:34] *** cncfanatics has quit IRC[11:12:50] *** trentster has joined #smartos[11:23:59] *** marsell has joined #smartos[11:35:06] *** Aram has joined #smartos[11:35:28] *** Aram has joined #smartos[11:39:42] *** Aram has quit IRC[11:41:00] *** Aram has joined #smartos[11:44:54] *** cncfanatics has joined #smartos[12:07:42] <izaki> Is there a pf implementation for Smartos? I am about to conclude there isn't but it won't hurt to ask...[12:08:46] <MerlinDMC> ipf?[12:09:18] <MerlinDMC> and the new (currently not documented) fwadm?[12:10:55] <izaki> MerlinDMC: I just found that and it made me feel like a n00b[12:11:00] *** d[^_^]b has quit IRC[12:11:03] <MerlinDMC> hm[12:11:03] <izaki> ;) Thanks![12:11:08] *** d[^_^]b has joined #smartos[12:11:32] <MerlinDMC> np? ;)[12:11:46] *** defa has joined #smartos[12:12:56] <Aram> aren't ipf and pf different?[12:13:46] *** defa has quit IRC[12:16:07] <arekinath> ipf and pf are considerably different. :)[12:19:03] <Licenser> at least a 33% difference[12:19:14] <Licenser> perhaps 50% depending on the POV[12:20:17] <izaki> pf and ipf look substantially different[12:20:37] <arekinath> but they are capable of similar things, yes[12:20:42] <arekinath> if you ignore pfsync. ;)[12:20:49] <arekinath> and pflow[12:20:52] <arekinath> and various other neat things[12:21:34] <arekinath> and don't mind writing 4x the config lines in a much less clear format to get the same job done *[12:21:44] <arekinath> * numbers may be made up[12:22:09] <Licenser> arekinath you don't like ipf I take it :P[12:22:47] <izaki> I only care about getting the job done... PF itself is senseless enough[12:23:37] <arekinath> Licenser: it's not that I don't like ipf, it's just that pf is better. :P[12:23:52] <Licenser> heh[12:25:45] <arekinath> izaki: I said that too, before I had to make my firewall do more than a few things[12:26:01] <arekinath> and apply to more than just one network of hosts[12:28:28] <arekinath> then the fact that pf is a simplified and clearer version of ipf starts to really make a difference to how nervous you are about hitting reload after editing rules[12:54:56] *** marsell has quit IRC[13:04:41] *** mamash has joined #smartos[13:26:09] *** avrntsv has joined #smartos[13:40:54] *** Aram has quit IRC[13:41:10] *** Aram has joined #smartos[13:46:45] *** mamash has left #smartos[13:47:34] *** mamash has joined #smartos[13:59:06] *** wdent has joined #smartos[14:01:02] *** jamesd has joined #smartos[14:56:41] *** mamash has left #smartos[15:17:18] *** avrntsv has quit IRC[15:19:20] *** avrntsv has joined #smartos[15:34:50] *** aksyn has joined #smartos[15:34:58] <aksyn> greetings earthlings[15:36:18] <norman> there is just one place with internet outside of the earth, and that is the ISS. are you on the ISS? :-)[15:36:32] <aksyn> norman: i'm not allowed to talk about such things*cch*[15:36:59] <aksyn> so i've just built a testbed boxen to run SmartOS on, and having trouble googling my way to converting/getting it to make the initial pool as a raid-z volume.. anyone have some wisdom to offer me?[15:39:02] <aksyn> i have 3 drives hooked up, c1t[0-2]d0.. but it's only taken the first one into the pool (as that was all I had on initial boot)[15:41:04] *** Aram has quit IRC[15:42:14] <aksyn> I can add the 2 remaining to a new raidz pool, that's easy enough.. but I want to include the first disk too.. possible?[15:43:03] <calmh> aksyn: i think you will need to "reinstall" with all disks present[15:43:17] <aksyn> calmh: will it offer me raidz as an option at that point, or is that the default?[15:44:03] <calmh> aksyn: it should be the default if you tell it to use all three disks in the initial setup[15:44:20] <aksyn> calmh: ahh, easy :) thanks for the pointer[15:44:49] <aksyn> calmh: how can i nuke the existing zones pool from the first disk?[15:44:56] <calmh> you might need to boot with destroy_zpools=true to get rid of the existing pol[15:45:05] <aksyn> k[15:46:33] *** wdent has quit IRC[15:51:59] <aksyn> raidz1 10.9T 6.01G 10.9T -[15:52:03] <aksyn> wooo :) thanks calmh[15:52:20] *** wdent has joined #smartos[16:11:36] *** mamash has joined #smartos[16:13:51] *** avrntsv has quit IRC[16:14:48] *** avrntsv has joined #smartos[16:26:10] *** aksyn_ has joined #smartos[16:26:11] *** aksyn has quit IRC[16:26:11] *** aksyn_ is now known as aksyn[16:33:18] *** dap has joined #smartos[16:38:08] *** dysinger has joined #smartos[16:41:37] *** mamash has left #smartos[16:52:36] *** mamash has joined #smartos[16:56:53] *** wdent has quit IRC[17:07:39] *** fsteinelX has joined #smartos[17:12:47] *** fsteinelX is now known as fsteinel[17:13:19] *** ryancnelson has joined #smartos[17:23:40] *** mamash has left #smartos[17:26:37] *** tonyarkles has joined #smartos[17:43:20] *** tonyarkles has quit IRC[17:59:28] *** ryancnelson has quit IRC[18:02:20] *** mamash has joined #smartos[18:10:10] *** tonyarkles has joined #smartos[18:13:39] *** wdent has joined #smartos[18:36:51] *** dysinger has quit IRC[18:54:42] *** kevinykchan has quit IRC[19:01:57] *** mamash has left #smartos[19:03:51] *** tonyarkles has quit IRC[19:12:02] *** wdent has quit IRC[19:12:45] *** wdent has joined #smartos[19:16:15] *** Aram has joined #smartos[19:16:20] *** Aram has joined #smartos[19:18:37] *** ryancnelson has joined #smartos[19:22:42] *** dysinger has joined #smartos[19:41:20] *** aksyn has joined #smartos[19:42:39] <aksyn> hi guys, is the e1000g driver generally considered stable? I have an odd thing happening on an Intel CT Desktop PCI-E NIC.. after some use, all connectivity stops, and netstat -i starts showing crazy numbers of input and output packets and errors… the really strange thing is they decrement!![19:46:49] *** jamesd has quit IRC[19:46:55] *** jamesd has joined #smartos[19:47:11] *** ryancnelson has left #smartos[19:52:22] <Triskelios> aksyn: that seems unusual. anything in dmesg about the interface?[19:52:36] <aksyn> bizarrely.. no, nothing![19:52:59] <aksyn> unplugging the LAN cable and back in doesn't reanimate it[19:53:14] *** avrntsv has quit IRC[19:53:33] <MerlinDMC> you have no networking on the GZ? ... but maybe zones are running fine on that host?[19:53:55] <aksyn> merlin: correct, affects GZ and all VMs (2x ubuntu) as well[19:54:45] <aksyn> any dtrace scripts out there to aid in triage?[19:54:51] <MerlinDMC> k ... my GZ died network side but running zones wee unaffected ... i also could connect from a zone to the GZ ... so that seems to be smth different ;)[19:55:04] <aksyn> merlin: but equally worrying :)[19:55:20] <MerlinDMC> hmm ... maybe :)[19:56:25] <aksyn> i'll keep a netstat -i rolling and reproduce it.. the numbers were so large it looked like the counters had overflowed[20:01:31] *** benbangert has quit IRC[20:08:11] *** notmatt has joined #smartos[20:08:52] *** benbangert has joined #smartos[20:10:59] *** dysinger has quit IRC[20:11:01] *** enmand_ has quit IRC[20:12:16] *** dysinger has joined #smartos[20:27:51] *** Raynos has quit IRC[20:34:52] *** Aram has quit IRC[20:35:25] *** notmatt has quit IRC[20:42:05] *** mamash has joined #smartos[20:57:11] *** dysinger has quit IRC[20:57:34] *** reiddraper_ has joined #smartos[20:57:45] *** dysinger has joined #smartos[20:59:40] *** Saskaloon has quit IRC[21:00:42] *** reiddraper has quit IRC[21:12:34] *** des2 has quit IRC[21:14:45] *** mamash has left #smartos[21:17:06] *** marsell has joined #smartos[21:20:01] *** des2 has joined #smartos[21:29:00] *** fsteinel has quit IRC[21:48:30] *** mamash has joined #smartos[21:51:48] *** Aram has joined #smartos[22:04:18] *** Saskaloon has joined #smartos[22:07:43] *** defa has joined #smartos[22:35:24] *** wblum has joined #smartos[22:54:53] *** mamash has left #smartos[22:59:09] *** enmand has joined #smartos[23:10:53] *** wblum has quit IRC[23:51:42] *** Aram has quit IRC[23:53:47] *** Raynos has joined #smartos[23:53:48] *** Aram has joined #smartos[23:56:49] *** jesser has joined #smartos