| nanvou | hello, I have a problem with upstart on Ubuntu 12.04 | 19:38 |
|---|---|---|
| nanvou | even though I submitted a "respaw" stanza, if the program terminate somehow, upstart won't reload it | 19:39 |
| nanvou | there's no script and only one "exec" stanzas... | 19:42 |
| nanvou | anyone ? | 19:54 |
| SpamapS | nanvou: respawn only respawns on abnormal exit | 20:01 |
| SpamapS | nanvou: make sure you read the respawn section carefully in 'man 5 init' | 20:01 |
| nanvou | process exits with 139 | 20:02 |
| SpamapS | nanvou: nothing in syslog about it? | 20:02 |
| nanvou | yes | 20:03 |
| nanvou | it syslog says it segfault | 20:03 |
| nanvou | which is normal | 20:03 |
| SpamapS | nanvou: segfault is also a reason to respawn (SIGSEGV) | 20:03 |
| nanvou | what isn't normal is that segfault is detected by system but upstart won't do anything about it and let the program stay stopped | 20:04 |
| SpamapS | nanvou: it is not marked 'task' right? and there's no 'normal exit' stanza? | 20:04 |
| nanvou | I ran the program manually. It does SEGFAUT with status code "139" | 20:04 |
| nanvou | SpamapS, http://paste.ubuntu.com/5623801/ | 20:05 |
| SpamapS | nanvou: indeed, upstart should be respawning that on segfault | 20:06 |
| SpamapS | nanvou: and syslog says, exactly? | 20:06 |
| nanvou | "kernel: [4526394.732639] segfaultest[25199]: segfault at fffffff7 ip bf9e085c sp bf9e085c error 4 | 20:06 |
| nanvou | " | 20:07 |
| SpamapS | thats kernel | 20:07 |
| SpamapS | nanvou: upstart would log as init: | 20:07 |
| SpamapS | nanvou: if the program is daemonizing before the segfault, you don't have 'expect fork', so upstart will lose track of it | 20:08 |
| nanvou | in fact it doesn't daemonize at all | 20:08 |
| SpamapS | nanvou: initctl log-priority info should show more of what init: is doing | 20:12 |
| nanvou | where should I wait for logs ? | 20:13 |
| SpamapS | syslog | 20:14 |
| nanvou | SpamapS, http://paste.ubuntu.com/5623839/ | 20:19 |
| SpamapS | nanvou: hah, so it did respawn | 20:20 |
| SpamapS | nanvou: eventually it gave up, respawning too fast | 20:20 |
| nanvou | ah ok | 20:21 |
| nanvou | so how do I imply some waiting time ? | 20:21 |
| nanvou | can I use multiple exec statement N | 20:25 |
| nanvou | *? | 20:25 |
| nanvou | like "exec sleep 10" ? | 20:25 |
| SpamapS | nanvou: respawn limit | 20:26 |
| SpamapS | nanvou: man 5 init explains it | 20:26 |
| nanvou | yes, but I don't want to limit the respawns themselves, more like the time between the respawns | 20:27 |
| SpamapS | nanvou: ah, post-start I think can achieve that. What is the point tho? | 20:27 |
| SpamapS | nanvou: if you have a program this unstable.... | 20:28 |
| nanvou | I'll make it better in the future | 20:28 |
| SpamapS | yeah but upstart is not the answer | 20:28 |
| nanvou | right now, I just need upstart to be able to respawn it not too fast | 20:29 |
| SpamapS | nanvou: just make a wrapper that always respawns it | 20:29 |
| nanvou | mmh | 20:29 |
| SpamapS | the upstart respawning is for *unexpected* fail | 20:29 |
| nanvou | yes, that sounds like a possibility | 20:29 |
| nanvou | SpamapS, thanks for the help | 20:49 |
| SpamapS | nanvou: any time | 20:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!