| crankharder | any way I can tell uptstart to load init files in a given path? | 14:40 |
|---|---|---|
| crankharder | include directive of sorts | 14:40 |
| jodh | crankharder: what are you trying to achieve? | 14:43 |
| crankharder | I want puppet to set up and configure a directory full of init scripts | 14:44 |
| crankharder | I can't tell puppet to monitor /etc/init since it'll delete all the scripts that puppet doesn't own | 14:44 |
| crankharder | so if upstart monitored /etc/init/puppet just like /etc/init we'd be good to go | 14:44 |
| jodh | crankharder: it does | 14:45 |
| crankharder | heh, hmm. ok | 14:45 |
| crankharder | jodh: https://gist.github.com/crankharder/4fa79577333ca89bd546 | 14:47 |
| crankharder | seems to not like that job | 14:47 |
| jodh | crankharder: What does 'init-checkconf <your_job>.conf' say? | 14:52 |
| crankharder | $ init-checkconf resque_worker_50.conf | 14:52 |
| crankharder | ERROR: file resque_worker_50.conf does not exist | 14:52 |
| jodh | crankharder: init-checkconf '/full/path/to/file.conf' | 14:53 |
| jodh | crankharder: your 'start on' should probably be 'start on runlevel [2345]' and your 'stop on' is currently wrong (unless you are emitting shutdown somewhere?) - it should be 'stop on runlevel [016]'. | 14:56 |
| jodh | crankharder: also, you don't need to do your own redirection/logging - upstart does it for you, logging all job output to /var/log/upstart/. | 14:57 |
| crankharder | had some perms issue, but now getting this: | 14:57 |
| crankharder | $ init-checkconf /etc/init/puppet/resque_worker_50.conf | 14:57 |
| crankharder | ERROR: failed to ask Upstart to check conf file | 14:57 |
| crankharder | /etc/init/puppet is 755 and the job is 744 | 14:57 |
| crankharder | jodh: okay, those points may be fair -- but I have the same job (named differently) in /etc/init and it works | 14:57 |
| jodh | crankharder: what distro version is this? | 14:58 |
| crankharder | ubuntu 12.04 | 14:58 |
| crankharder | I get this too: | 14:59 |
| crankharder | $ init-checkconf /etc/init/resque_worker_5.conf | 14:59 |
| crankharder | ERROR: failed to ask Upstart to check conf file | 14:59 |
| crankharder | for someone not in my nested path | 14:59 |
| crankharder | something | 14:59 |
| jodh | crankharder: copy the file to /tmp, ensure it is readable, then 'init-checkconf -d /tmp/foo.conf' | 15:01 |
| jodh | crankharder: if that still fails, I'm guessing you have a dbus issue - is dbus-launch available? | 15:01 |
| jodh | crankharder: your job can also use use 'setuid', 'setgid' and 'chdir' to simplify the main job to be just 'exec rake environment resque:work QUEUE=high,medium RAILS_ENV=production PIDFILE=/tmp/resque_worker_50.pid VERBOSE=1' I think. | 15:04 |
| jodh | crankharder: putting pid files in /tmp looks like a bad idea btw. | 15:04 |
| crankharder | yea, looks like dbus issues. no, i dont have dbus-launch | 15:06 |
| crankharder | output here: https://gist.github.com/crankharder/1ee1e4cb50cc8353a3e4 | 15:08 |
| crankharder | it's a bit weird init-checkconf doesn't work on scripts in /etc/init/ ? -- maybe this isn't the issue? | 15:09 |
| jodh | crankharder: maybe copy your job to a desktop system with a full dbus setup and run init-checkconf on it there. I wonder if you have any odd whitespace chars in that job - copied from windows maybe? | 15:20 |
| crankharder | so, testing on a more barebones ubuntu 12.04 isntance I have w/o my jobs in it... installing dbus-launch now | 15:21 |
| jodh | crankharder: another way to check the job without using init-checkconf is: sudo initctl log-priority debug;sudo touch /etc/init/foo.conf;sudo initctl log-priority message' and look at dmesg/syslog to see what messages upstart logged about foo. | 15:21 |
| crankharder | $ init-checkconf /etc/init/acpid.conf │ | 15:22 |
| crankharder | ERROR: failed to ask Upstart to check conf file | 15:22 |
| crankharder | -d shows samee waiting for dbus issues | 15:22 |
| jodh | crankharder: init-checkconf won't work unless dbus-launch works. | 15:23 |
| crankharder | not really familliar with dbus launch. define works? https://gist.github.com/crankharder/53351450db600a15252e | 15:24 |
| jodh | crankharder: let's forget init-checkconf - I've run it locally and your job as pasted is syntactically valid. I've given you another option above to syntax check anyway. I suspect the job on disk is not the same as pasted - it might contain '\r's or similar if you copied it from a windows machine maybe? | 15:26 |
| jodh | crankharder: try the technique above and see what upstart logs to the system log - it will tell you where it is failing to parse the job. | 15:27 |
| crankharder | okay, so i'm on a new system w/o my custom jobs and I still see the same issue. It's not related to my custom job. | 15:29 |
| jodh | crankharder: which issue? the job or init-checkconf? | 15:29 |
| crankharder | https://gist.github.com/crankharder/cd04bcb4d80259ba58d3 | 15:29 |
| jodh | crankharder: 'sudo status puppet/tty6' | 15:29 |
| crankharder | okay, that works | 15:30 |
| crankharder | unfortunately I see nothing in syslog or dmesg from this: sudo initctl log-priority debug ; sudo touch /etc/init/cron.conf ; sudo initctl log-priority message | 15:32 |
| jodh | crankharder: are you by chance running in a cloud environment that uses overlayfs? | 15:33 |
| crankharder | i'm running ubuntu in virtualbox... but i dont know what overlayfs is, so certainly didn't set tha tup | 15:33 |
| jodh | crankharder: mount|grep overlay | 15:34 |
| crankharder | nothing | 15:34 |
| crankharder | okay, I removed all my custom init scripts and moved tty6 into this puppet path. giving my unknown job puppet/tty6 on this system (but it worked on the more barebones one) | 15:38 |
| jodh | crankharder: this makes no sense - your system sounds broken. | 15:38 |
| crankharder | /etc/init/puppet is 755 on both /etc/init/puppet/tty6.conf is 744 on both | 15:39 |
| crankharder | I can test this at different points as puppet sets up the system to see if that helps | 15:41 |
| crankharder | jodh: thanks for the help -- i'll post here if I can track down what change makes this break | 15:43 |
| jodh | crankharder: you said this job runs fine in /etc/init/ but not in /etc/init/puppet/ - is that now resolved now you know how to start a job in a sub-directory? | 15:44 |
| crankharder | the subdirectory thing works on a box that hasn't had all my puppet configs (lots of them) applied. same version of ubuntu tho | 15:45 |
| crankharder | so I assume it's something puppet is configuring that's causing this | 15:45 |
| jodh | crankharder: I'd certainly make sure you have a pristine server environment available (ie no puppet) for testing the basic upstart jobs. | 15:48 |
| CharlieSu | I've got an UpStart task that I'd like to print a message to the console once run.. 'start job' should produce something on the users screen.. How can I do this? I've tried 'console output' | 17:04 |
| xnox | CharlieSu: at boot "console output" will print to the active console / boot messages. If run interractively it will print where-ever it's run. | 17:07 |
| CharlieSu | xnox: doesn't seem to be doing that for me.. I just use "logger 'message'" right? | 17:31 |
| xnox | CharlieSu: no, just do "echo" | 17:38 |
| xnox | logger will go whever logger logs, e.g. syslog. | 17:38 |
| xnox | CharlieSu: "console " stanza controls where job's stdout and stderr is redirected, default is "log" which is under /var/log/upstart/, "output" is to normal stdout and etc. | 17:39 |
| === JanC_ is now known as JanC | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!