| === bradleyayers_ is now known as bradleyayers | ||
| muzzol | hi | 11:49 |
|---|---|---|
| muzzol | is there any variable that holds the action that triggers it? (start, stop, restart) | 11:50 |
| muzzol | i want to do something like this: http://pastebin.com/XFJJsceu | 11:51 |
| muzzol | is that possible? | 11:51 |
| lpapp | jodh: hey, I do not understand what upstart does, but when I run the command manually, the "daemon" works as expected unlike with an upstart job :( | 16:49 |
| jodh | lpapp: where does $ACTION come from? | 16:50 |
| jodh | lpapp: have you looked at http://upstart.ubuntu.com/cookbook/ ? | 16:51 |
| jodh | lpapp: your job looks good aside from that variable which isn't being set, so the script will presumably fail. See also http://upstart.ubuntu.com/cookbook/#obtaining-a-log-of-a-script-section | 16:52 |
| djszapi | jodh: do you have a clue what can cause that ? | 16:56 |
| djszapi | http://paste.kde.org/483674/ -> this is my upstart job | 16:57 |
| doki_pen | what's the best way to kick off a proc as an unprivileged user? | 17:12 |
| djszapi | init-checkconf /etc/init/foobar.conf | 17:23 |
| djszapi | ERROR: failed to ask Upstart to check conf file | 17:23 |
| djszapi | what is this, SpamapS ideas ? | 17:23 |
| djszapi | slangasek: same issue again | 17:25 |
| djszapi | the exec line works, if I run manually. | 17:26 |
| djszapi | but if I use start foobar, it does not. | 17:26 |
| slangasek | oh, really? | 17:26 |
| slangasek | so even running 'start foobar' fails? | 17:26 |
| djszapi | nope | 17:27 |
| djszapi | funky thing is that, the ps aux | 17:27 |
| djszapi | output shows the process execed :D | 17:27 |
| djszapi | but the daemon does not work as expected. | 17:27 |
| slangasek | djszapi: is the code for the daemon open? | 17:27 |
| slangasek | also, can you try modifying the job to run /usr/bin/foobar under strace, and capture the syscall output? | 17:28 |
| djszapi | slangasek: first step would be to get the daemon output logged into somewhere. | 17:28 |
| djszapi | no, the daemon is a commercial stuff | 17:29 |
| djszapi | cannot disclose, sorry. | 17:29 |
| slangasek | ok | 17:29 |
| djszapi | but that helps, if I know how to log the output. | 17:29 |
| slangasek | daemon output logged> what version of Ubuntu are you running again? | 17:29 |
| slangasek | the upstart from 12.04 has logging support | 17:29 |
| djszapi | Linux linaro-ubuntu-desktop 3.1.1-26-linaro-lt-omap #26~lt~ci~20120325001352+1332635991~4f6ec49b-Ubuntu SMP PREEMPT armv7l armv7l armv7l GNU/Linux | 17:30 |
| djszapi | 1.3-0ubuntu12~linaro2 | 17:31 |
| slangasek | if you're using 12.04, or if you can upgrade upstart to that version, you'll get a /var/log/upstart/foobar.log containing all the output | 17:31 |
| slangasek | otherwise, you can use shell redirection directly in the upstart job | 17:31 |
| slangasek | exec /usr/bin/foobar receive /dev/ttyUSB0 38400 > /var/log/my-log 2>&1 | 17:31 |
| slangasek | ^^ should work | 17:31 |
| slangasek | if it *doesn't* work, you just need to wrap the exec with a script / end script | 17:32 |
| djszapi | let us see if that redirection works. | 17:33 |
| djszapi | slangasek: foudn the issue. The log helped a lot, thanks. | 17:59 |
| djszapi | it is now working :) | 17:59 |
| slangasek | ok | 17:59 |
| gchristensen | hi, my upstart script seems to fail to start when the system is starting, even though I've given it a very relaxed respawn limit, and even put a 3 second sleep in the script. here it is: https://gist.github.com/96512f0ef329d5b022e4 | 21:56 |
| slangasek | gchristensen: a) your start and stop conditions are identical, which is certainly an error; b) 'mysql' is not a standard upstart event, so unless you've written something that outputs it, you'll never satisfy the start condition | 22:00 |
| gchristensen | I see | 22:00 |
| slangasek | you probably want 'start on started mysql' | 22:00 |
| slangasek | and 'stop on stopping mysql' | 22:00 |
| JoeJulian | If an upstart script has to wait for the network to be started, what's the string it needs to WAIT_FOR? | 22:01 |
| gchristensen | slangasek: that might do it. thank you for your help, I'll read the manual closer | 22:02 |
| slangasek | your 'sleep 3' should be unnecessary, btw; the mysql upstart job should, like all upstart jobs, have already been written such that the mysql server is up and ready to receive queries before emitting 'started' | 22:02 |
| gchristensen | slangasek: that was a last-ditch attempt | 22:02 |
| slangasek | JoeJulian: unless there's a specific reason that your service needs to start earlier, best practice here is 'start on runlevel [2345]' | 22:03 |
| JoeJulian | So the network should be started already at that point? | 22:04 |
| slangasek | JoeJulian: definitely | 22:04 |
| JoeJulian | The problem seems to be with semiosis' upstart scripts for glusterfs. One user has encountered an issue with resolving hostnames while glusterd is starting. I presumed this was due to the network not being started. | 22:05 |
| slangasek | JoeJulian: the exception is if your network is controlled by network-manager, in which case there's no guaranteed synchronization event for the network being up | 22:05 |
| JoeJulian | Does debian use network-manager? (not a debian user myself) | 22:05 |
| slangasek | JoeJulian: Debian supports network-manager... it would not be used by default on a server, be it Debian or Ubuntu | 22:06 |
| slangasek | and Debian also doesn't use upstart | 22:07 |
| slangasek | (currently) | 22:07 |
| JoeJulian | Ok. Thank you. I'm not very good at supporting non-rpm based distros. :D | 22:08 |
| slangasek | :) | 22:09 |
| slangasek | so are the upstart jobs that you're concerned about different than the ones in the glusterfs-server package shipped in Ubuntu? | 22:09 |
| slangasek | (which does indeed show 'start on runlevel [...]' | 22:10 |
| slangasek | ) | 22:10 |
| JoeJulian | I /think/ that those are, indeed, the ones that semiosis wrote. | 22:10 |
| JoeJulian | And I got two users mixed up. It is ubuntu. | 22:11 |
| slangasek | right - that's much more plausible :) | 22:12 |
| slangasek | so yeah, if it was a server install, the existing job should already be correct | 22:12 |
| slangasek | if someone's doing this on a desktop install, there could be problems, depending on the configuration of network-manager | 22:12 |
| JoeJulian | This guy is a former windows admin. I think you're on to something with the networkmanager idea. | 22:12 |
| gchristensen | slangasek: what "start on" output should I be listening to for something that expects networking and a writable filesystem? ie: a generally usable machine | 22:13 |
| slangasek | gchristensen: that's precisely the 'start on runlevel [2345]' I mentioned to JoeJulian | 22:13 |
| JoeJulian | Looks like you need a bot with an automatic notice. :D | 22:13 |
| JoeJulian | hehe | 22:13 |
| slangasek | if you look in /etc/init/rc-sysinit.conf, you'll see this defined in terms of 'filesystem and static-network-up' | 22:14 |
| gchristensen | I see | 22:14 |
| gchristensen | thank you | 22:15 |
| JoeJulian | "Desktop 32bit. Stock image"! Thanks slangasek. Good catch. | 22:24 |
| slangasek | ok | 22:24 |
| slangasek | too bad that there's no upstart event to key on in that case | 22:25 |
| slangasek | but yeah, if the user can switch to static network config via /etc/network/interfaces, it'll work around this problem | 22:25 |
| JoeJulian | It's supposed to be a server anyway. I'll just make him reinstall. | 22:26 |
| slangasek | :) | 22:27 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!