Back to Home / #linode / 2006 / 09 / Prev Day | Next Day
#linode IRC Logs for 2006-09-15

---Logopened Fri Sep 15 00:00:47 2006
00:12|-|fo0bar [] has quit [Quit: leaving]
00:12|-|fo0bar [] has joined #linode
00:14|-|fo0bar [] has quit []
00:18|-|spr [] has quit [Quit: This computer has gone to sleep]
00:19|-|fo0bar [] has joined #linode
01:55|-|SpaceHobo [] has quit [Ping timeout: 480 seconds]
02:03|-|Netsplit <-> quits: darkbeholder, lucca, Dreamr3, AndyHat, Skule, guinea-pig, Beirdo
02:03|-|Netsplit over, joins: darkbeholder
02:03|-|Netsplit over, joins: Dreamr3
02:03|-|Netsplit over, joins: AndyHat
02:03|-|Netsplit over, joins: Beirdo
02:03|-|Netsplit over, joins: Skule
02:03|-|Netsplit over, joins: guinea-pig
02:03|-|Netsplit over, joins: lucca
02:05|-|fapestniegd [] has quit [Quit: CGI:IRC (Session timeout)]
02:14|-|internat [] has joined #linode
02:31|-|lucca [] has quit [Remote host closed the connection]
02:32|-|lucca [] has joined #linode
02:46|-|Eman [] has quit [Read error: Connection reset by peer]
02:48|-|Eman [] has joined #linode
02:58|-|rko [] has quit [Quit: Trillian (]
03:10|-|SpaceHobo [~spacehobo@] has joined #linode
04:15|-|SpaceHobo [~spacehobo@] has quit [Ping timeout: 480 seconds]
04:20|-|SpaceHobo [~spacehobo@] has joined #linode
05:02|-|M-guy changed nick to Aries
05:02|-|Aries changed nick to M-guy
07:19|-|linville [] has joined #linode
07:55|-|efudd2 [] has joined #linode
07:55<efudd2>So. Caker, mike?
08:08|-|spr [~spr@] has joined #linode
08:10|-|efudd [] has joined #linode
08:10[~]efudd grumbles
08:12<efudd>I guess host29 rebooted.. my linode didn't restart with a bogus error of no root partition found... panic'd twice restarting...
08:13<efudd>now it seems to run.
08:13<efudd>this after 100+ days of uptime on my node....
08:14|-|efudd2 [] has quit [Quit: CGI:IRC]
08:15<bendy24>i feel for you man
08:16|-|Dreamr_3 [] has joined #linode
08:16<efudd>.. and now apparently i can't auth against my cyrus :)
08:16<efudd>Kernel panic - not syncing: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
08:16<bendy24>my host went down last week
08:16<bendy24>forever borked, indeed
08:17<bendy24>oh wait
08:17<bendy24>thats broket
08:21|-|Dreamr3 [] has quit [Ping timeout: 480 seconds]
08:34|-|Eman [] has quit [Server closed connection]
10:00|-|BitterKvandivo changed nick to kvandivo_stillBitter
10:47|-|FireSlash [~FireSlash@] has joined #linode
11:19<@caker>[root@host29 root]# uptime
11:19<@caker> 12:19:21 up 453 days, 1:46, 2 users, load average: 0.69, 0.63, 0.56
11:23|-|internat [] has quit [Quit: This computer has gone to sleep]
11:26|-|SpaceHobo [~spacehobo@] has quit [Ping timeout: 480 seconds]
11:36|-|jekil [~alessandr@] has joined #linode
11:36|-|FireSlash [~FireSlash@] has quit [Quit: Leaving]
11:46|-|benson [] has joined #linode
11:48|-|benson7 [] has joined #linode
11:51|-|benson [] has quit [Quit: CGI:IRC (Ping timeout)]
12:00|-|afv-13 [] has joined #linode
12:01|-|SpaceHobo [] has joined #linode
12:06<efudd>caker, odd.
12:06<efudd>wonder what the hell happened then.
12:07[~]efudd looks closer
12:07<efudd>i ass-u-me'd it was the host.
12:09<efudd>neat. My host knows nothing of the time between ~midnight and 8AM when i rebooted it.
12:09<efudd>and s/reboot/boot/
12:10<efudd>Sep 15 13:10:20 [imaps] TLS server engine: cannot load CA data
12:11[~]efudd ponders.
12:11<efudd>anything change with the kernels re: tls?
12:18<iggy>different TLS
12:20<npmr>efudd, the most recent 2.6 uml kernels support nptl
12:21<npmr>efudd, i had a similar unexplained shutdown a few weeks ago
12:22<npmr>turned out to be a fatal kernel oops in the uml kernel
12:26|-|benson7 [] has quit [Quit: benson7]
12:41|-|Dreamr3 [] has joined #linode
12:45|-|afv-13_ [] has joined #linode
12:45|-|afv-13 [] has quit [Read error: Connection reset by peer]
12:46|-|Dreamr_3 [] has quit [Ping timeout: 480 seconds]
13:02<@caker>efudd: yeah, too bad you didn't check Lish's logview before booting those few times...
13:17<efudd>ayup. fair enough. too bad.
13:27<efudd>motherfucker. my cyrus is all fucked up since this.
13:34<efudd>weird; had to -re-add my users to fucking sasl.
13:43<@caker> <-- jdike is working on SMP UML
13:51<afv-13_>that would be great
13:52<afv-13_>caker: have there been any improvements in the world of Xen?
14:02<Ciaran>Hmm. Out of interest, how often are's and's nameservers updated with reverse DNS data for our Linodes?
14:18|-|afv-13_ [] has quit [Read error: Connection reset by peer]
14:23|-|afv-13 [] has joined #linode
15:26<taupehat>what the heck was that?
15:26[~]taupehat is experiencing networking burpage on his linode
15:40|-|rko [] has joined #linode
15:41<@caker>Ciaran: I think HE pulls them in twice a day
15:42<@caker>afv-13: not on our end (although our end has been fairly bug free, except for some host-restart-all-xenodes issues)...
15:42<@caker>About two weeks ago, Xensource started selling XenEnterprise, so I suppose they at least think it's production ready
16:09|-|internat [] has joined #linode
16:10|-|jekil [~alessandr@] has quit [Ping timeout: 480 seconds]
16:13|-|internat [] has quit []
16:43|-|linville [] has quit [Quit: Leaving]
17:02|-|Eman [] has joined #linode
18:00<guinea-pig>caker: so where does linode stand with Xen currently, then?
18:00<guinea-pig>(I've been so out of touch :( )
18:01<guinea-pig>i mean, that's all i do now... stop in every couple months and ask the same question :P
18:05<klaruz>i'm wondering the same thing
18:05<kvandivo_stillBitter>a true guinea-pig would be here all the time, patiently waiting first in line to try out all the new stuff
18:05<guinea-pig>a true guinea-pig would be dead for trying out something first
18:06<guinea-pig>a true guinea-pig cares for nothing more than fresh letuces
18:06[~]guinea-pig nibbles his fresh lettuces
19:04<@caker>guinea-pig: well, the beta was successful in the sense that it helped us eliminate most of the bugs on our end, but we decided to wait a while longer for Xen to mature (read: not crash)
19:05<@caker>every so often, I pull in the latest Xen changes, but we've still seen crashes every few months
19:06<@caker>The other thing is Xen lacks a method to do disk I/O QoS, other than to adjust CFQ's ioprio values (which still don't have hard-limits for thrashing nodes) or to pause the offending node
19:07<@caker>The "other" Xen-only provider even stopped putting new hosts online, and publically stated that Xen wasn't production ready (ouch)
19:08<@caker>so we're just in a holding pattern until Xen is ready... The good news is that new hosts we've placed online are already LVM-backed, so essentially just a reboot for those hosts from uml<->xen
19:10<klaruz>good to know, sounds like something that would be good to post on the forum
19:16<guinea-pig>fun fun
19:17<fo0bar>caker: do you know if file-backed images are workable in xen yet? I know for awhile it was very flaky (which is probably why you went with LVM), but file-backed ISO images (IE, finnix) have worked pretty well in my experience with xen
19:18<@caker>fo0bar: from what I've read it works again ... the only thing I've read about them is some guy doing perf testing compared to LVM/partition backed images, and why there was a perf difference (host caching)
19:18<guinea-pig>well, always curious, and a bit of a guinea-pig... but i don't really care one way or another. with all the memory upgrades... i think i'm due for downgrading my linode anyway :P
19:18<@caker>other than that, I haven't followed the Xen ML much of late
19:19<fo0bar>caker: ahh. well, if you have a good management system for keeping all the LVs sane, LVM is probably going to be the better choice anyway
19:19<guinea-pig>heh, it's been two months since i've touched swap
19:20<@caker>It would have made life much easier to just stick with file images (our LVM implementation is done, anyway), but it was more for performance than anything else (although at the time, file backed didn't work)
20:26[~]efudd blinks
20:26<efudd>still wish i'd know what happened to this damned box :/
20:34<guinea-pig>you said it
20:34<guinea-pig>it's a damned box.
20:34<guinea-pig>i think i'm gonna build a hobbit hole
20:37<guinea-pig>i found a few people trying to do so... not liking any of their layouts, however
20:37<efudd>er, I'm being dumb here. What?
20:38<guinea-pig>hobbit hole.
20:38<guinea-pig>like a hobbit lives in
20:38<guinea-pig>but ... um human sized
20:38<efudd>Er, Ok. That's what I first thought...
20:38<efudd>That's, hrm, Neat? :)
20:38<efudd>(insert tone of slight sarcasm and pointed-mumbling-laughter).
20:39<guinea-pig>someone put up a donation-request site for building his own, 2 years ago. so far he's recieved $535
20:39<efudd>Well, that's one way to get money. It's just like begging on the side of the road.
20:39<guinea-pig>i bet i could double that for my own in a month :P
20:39<efudd>in fact
20:39[~]efudd goes to register
20:39<efudd>damn. beast.
20:39<efudd>beat too.
20:39<guinea-pig>there's no place like hom
20:39<guinea-pig>there's no place like hom
20:39<efudd>No match for domain "HONGRYNHOMLESS.COM".
20:39<efudd>that's very tempted.
20:40<efudd>there's a guy on the way to UpperClassTown that's got a sign with that on it.
20:40<efudd>I think you mean -ing.
20:40[~]efudd quotes terry pratchett
20:40<guinea-pig>don't question my mean-ing
20:40<efudd>-ing ok!
20:40<guinea-pig>f-ing right
20:41<guinea-pig>and you can take that to the bank!
20:41<guinea-pig>(won't get anything for it, though)
20:50<erikh>that wii is pretty
20:50<erikh>but personally, I know all I'll get with it is the nintendo titles.
21:19<Battousai>also, ubisoft
22:19|-|rjp [] has joined #linode
22:20<rjp>server3 seems to be down, or maybe it's the planet having trouble
22:20<rjp>er, host3
22:22<rjp>doing a traceroute from a HE linode, the last host to report is (
22:22<guinea-pig>i'm getting through just fine from home
22:22<guinea-pig>and from my HE linode
22:22<guinea-pig>to host3, that is
22:23<guinea-pig>are you tracing to the host or your linode? :P
22:23<rjp>hmm, just logged in to my home system on comcast, and a different host is last
22:24<rjp>my host, but a SSH attempt to LISH failed - maybe i'll try again
22:24<rjp>hmm... must have been rebooted, this time LISH worked
22:25<rjp>tryinb to boot now
22:25<rjp>not responding :/
22:26<rjp>there we go
22:26<rjp>looks like either host3 or the linode itself crashed :(
22:27<rjp>"check forced" on the ext2 filesystem I use for BT seeds
22:28<rjp>wasn't even *running* bt when the crash happened
22:30<rjp>finally... coming up. looking good now
22:30|-|rjp [] has quit [Quit: Bus error - passengers dumped]
22:59|-|VS_ChanLog [] has left #linode [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #linode
23:03|-|efuddSigh [] has joined #linode
23:03<efuddSigh>So, caker...
23:03<efuddSigh>any idea why my node is suddenly panic'ing?
23:03<efuddSigh>Kernel panic - not syncing: switch_mm_skas - PTRACE_SWITCH_MM failed, errno = 9
23:03<efuddSigh>after > 100 days uptime ....
23:04<efuddSigh>and heck, no panic's at all that I can remember in 3 years in the UML.
23:09|-|efudd [] has quit [Ping timeout: 480 seconds]
23:10|-|efudd [] has joined #linode
23:10[~]efudd grumbles
23:12<@caker>!errno 9
23:12<@linbot>caker: EBADF (#9): Bad file descriptor
23:14<@caker>efudd: hmm .. what was the kernel with lotsa uptime?
23:14<@caker>I've been wanting to build a new kernel but have been waiting until 2.6.18 goes gold
23:15<efudd>it was /probably/ 2.6.17-rc2-linode19
23:15<efudd>but -that- panic'd when i first restarted my node this AM.
23:15<efudd>that's the whole reason i changed to "latest 2.6"
23:15<@caker>best thing I can suggest is to revert to an older kernel
23:15<@caker>Label: 2.6.17-rc2-linode19 [testing]
23:15<@caker>Based on: Linux 2.6.17 (changelog)
23:15<efudd>same panic reason for it...
23:15<@caker>Last updated: Apr 26,2006
23:16<efudd>Sure, I'll go back to it, but again, it panic'd on boot also this morning.
23:16<@caker>well, maybe farther back then? .. could you deal with 2.4?
23:16<efudd>That is a completely bunk suggestion.
23:16<@caker>wow .. haven't updated 2.4 in over a year
23:17<efudd>now I know how my customers feel when TSEs tell them to upgrade without a reason. heh.
23:18<efudd>Hmmm. you ever figure out why sshd on host29 was fonky the other day and was fixed by a restart?
23:19<efudd>I'm just wondering if the base kernel is getting sompn strange in memory after a long uptime.
23:19<@mikegrb>the sshd thing is pretty common
23:19<@caker>well, that seems to happen often (root cause is syslog getting stuck)
23:19<@caker>but yeah, host could use an update
23:19<@caker>nothing odd in dmesg/other logs on the host
23:20<efudd>Yeah, all I get is that one line on console...
23:20<efudd>no other indications of the reason or even the event...
23:20<@caker>maybe get rid of /lib/tls, if you can -- just to be sure it's not a NPTL bug of some kind
23:22<efudd>it's happened before; according to google other folk were having this same panic back in october of last year i guess it was.
23:22<efudd>oh, way back in 2003. heh.
23:22<efudd>but for errno3.
23:22<efudd>!errno 3
23:22<@linbot>efudd: ESRCH (#3): No such process
23:23<@caker>efudd: if you reboot once more, I added "stderr=1" which *might* provide more output next time (I'm doubtful, however)
23:24|-|efudd [] has quit [Quit: leaving]
23:26<efuddSigh>jdike is pretty darned smart....
23:26<guinea-pig>what ever gives you that impression?
23:27<efuddSigh>caker, what's the current pidcount on the host for 29?
23:27<efuddSigh>high pid count I should specify...
23:28|-|efudd [] has joined #linode
23:29<encode>i have a question about gcc 4.0 - trying to install on debian sarge stable
23:29<encode>but its not there, so i tried the unstable sources - it has 4.1 and 4.0
23:30<encode>so i installed 4.0, but i dont get the executable gcc
23:30<encode>so i did an apt-get install gcc and it wants to install 4.1
23:30<encode>any ideas?
23:30<guinea-pig>the gcc package provides symlinks to the current default-gcc-compiler version
23:30<guinea-pig>on testing, it's 4.1
23:31<encode>hmm ok
23:31<@caker># cat /proc/sys/kernel/pid_max
23:31<guinea-pig>why 4.0 specifically?
23:32<encode>cc -I. -O4 -g -Wall -DBUILDING_TONEZONE -DSTANDALONE_ZAPATA -DZAPTEL_CONFIG=\"/etc/zaptel.conf\" -DHOTPLUG_FIRMWARE -c -o gendigits.o gendigits.c
23:32<encode>make: cc: Command not found
23:32<encode>how can i fix that?
23:33<encode>why 4.0? cos im trying to install asterisk on this server for someone i know, and they have this modified debian distro which had its kernel compiled with 4.0
23:33<guinea-pig>err, so?
23:34<encode>i need to build a kernel module
23:34<guinea-pig>ah, right
23:34<encode>do i need to symlink /usr/bin/cc to gcc-4.0 too?
23:34<guinea-pig>cc should be an alternative... provided by the gcc package, as well
23:34<guinea-pig>again, debian style
23:35<guinea-pig>caker: shouldn't the host's pid_max be sum(linode's pid_max) ?
23:35<@caker>guinea-pig: maybe back in TT mode days, yeah
23:35<encode>but i cant use the gcc package, because it will only install 4.1
23:36<efudd>Eh, almost seems like it'd require a cpu register error to cause ptrace to fail in PTRACE_SWITCH_MM
23:36<guinea-pig>encode: 4.0 vs 4.1 isn't like 2.95 vs 3.x transition
23:36<encode>but if i compile with 4.1 i get erros about magic numbers
23:36<guinea-pig>i'd be willing to bet you could get away with it
23:36<guinea-pig>ooh magic.
23:37<efudd>caker, I'm sure the answer is no, but just out of diligence, the host didn't print out the current stack/CPU registers when a node does this bychance?
23:37<encode>well compiling with 4.0 hasnt helped me - i just get different errors
23:37<encode>zaptel: disagrees about version of symbol copy_from_user
23:37<encode>zaptel: Unknown symbol copy_from_user
23:38<@caker>efudd: nope, no output unless something awful happens on the host
23:38<efuddSigh>uml's implementation is pretty damned neat as I look into it.
23:38<efuddSigh>spiffy tricks.
23:38<efuddSigh>toss a coupla pages of context on the stack
23:38<efuddSigh>and profit.
---Logclosed Sat Sep 16 00:00:35 2006