Zeqadious | does "start on stopped job ok" still work? | 18:58 |
---|---|---|
Zeqadious | I ask because it doesn't seem to here | 18:58 |
sadmac2 | Zeqadious: version? | 18:58 |
Zeqadious | 0.5.1 | 18:58 |
sadmac2 | Zeqadious: start on stopped job RESULT=ok | 18:59 |
Zeqadious | sweet | 18:59 |
Zeqadious | whats the other RESULT? | 18:59 |
Zeqadious | RESULT=failed? | 18:59 |
Keybuk | surprisingly | 18:59 |
sadmac2 | Zeqadious: although I'd think that would still work... maybe its changed a bit. There's a way | 18:59 |
sadmac2 | other result was failed I think | 19:00 |
Zeqadious | ok | 19:00 |
Keybuk | you can also just shortcut RESULT=failed and just check EXIT_STATUS=* or something | 19:00 |
Zeqadious | I removed all the ok's | 19:00 |
Zeqadious | i'll put them back now | 19:00 |
Keybuk | start on stopped job EXIT_SIGNAL=11 etc. | 19:00 |
* sadmac2 tries to imagine the task you want to run on segfault and nowhere else | 19:00 | |
Keybuk | start on stopped EXIT_SIGNAL=11 | 19:01 |
Keybuk | exec apport-auto-file $JOB | 19:01 |
sadmac2 | Keybuk: wouldn't you want SIGABRT in there too? | 19:02 |
sadmac2 | also SIGILL, SIGFPE, possibly SIGPIPE | 19:03 |
Zeqadious | hmmm "start on started udev-daemon RESULT=ok" should work right? | 19:35 |
sadmac2 | it should | 19:36 |
sadmac2 | unless it became OK or something like that. | 19:36 |
* sadmac2 doesn't have the code in front of him to check on these things | 19:36 | |
Zeqadious | i'll try that.. | 19:36 |
Zeqadious | fsck comes after that, but its not starting for some reason | 19:37 |
sadmac2 | Zeqadious: you could do a start on started udev-daemon and then in the job definition do an export > /tmp/somefile and then look at somefile and see what its getting set to | 19:37 |
Zeqadious | export what though? | 19:38 |
Zeqadious | export RESULT > /tmp/$job.result ? | 19:39 |
Zeqadious | oops | 19:39 |
Zeqadious | $RESULT | 19:39 |
sadmac2 | Zeqadious: I meant set | 19:40 |
sadmac2 | Zeqadious: no arguments | 19:41 |
* sadmac2 begins to realize how many ways unix has to list all the environment variables | 19:41 | |
Zeqadious | env | 19:41 |
sadmac2 | set does it too | 19:41 |
Zeqadious | really? | 19:41 |
sadmac2 | export I think does a superset of it | 19:41 |
sadmac2 | yes | 19:42 |
Zeqadious | never tried that | 19:42 |
Zeqadious | so it does | 19:42 |
Zeqadious | i'll check the result now | 19:43 |
Zeqadious | lol, forgot that before fsck the fs is ro | 19:44 |
sadmac2 | hmm. just echo it then. stdout should display it | 19:45 |
Zeqadious | yep | 19:45 |
Zeqadious | well it has no result really | 19:47 |
sadmac2 | the environment should be small | 19:47 |
Zeqadious | at least there is no variable called RESULT | 19:47 |
Zeqadious | it is | 19:47 |
Zeqadious | UPSTART_EVENTS=started | 19:47 |
Zeqadious | closest thing i can find | 19:47 |
sadmac2 | odd | 19:47 |
Zeqadious | is this the correct way to use the expressions with the start on stanza http://zeqadious.homelinux.net/junkbin/upstart-0.5.1-jobs/sysinit/save-dmesg ? | 20:06 |
sadmac2 | Zeqadious: maybe... what do you think that will do?' | 20:09 |
Zeqadious | I think it should start when all conditions inside ( ) are met | 20:09 |
Zeqadious | but i may not need the ( ) | 20:10 |
sadmac2 | Zeqadious: they aren't conditions. They're events. It will start when all the things inside () /happen/ | 20:10 |
sadmac2 | and you need either the \s or the ()s, not both | 20:10 |
Zeqadious | oh? | 20:10 |
sadmac2 | oh. | 20:10 |
Zeqadious | it won't tell me that there are unknown stanzas when it sees and started sysinit/cleanup RESULT=ok on a line all by itself? | 20:11 |
sadmac2 | Zeqadious: it assumes the line continues until the closing parenthesis | 20:12 |
Zeqadious | thats great | 20:12 |
Zeqadious | looks a lot nicer | 20:12 |
Zeqadious | well that doesn't seem to work right. that job never starts and afaik those other jobs are all RESULT=ok | 20:38 |
sadmac2 | Zeqadious: are all of them started in the first place? | 20:42 |
Zeqadious | yes | 20:42 |
sadmac2 | ok. that's the usual reason these things break... | 20:42 |
Zeqadious | hardest part i'm finding with these upstart scripts is trying to make them robust enough to deal possible breaks | 20:45 |
sadmac2 | example? | 20:45 |
Zeqadious | i'm having issues with fsck script for example. its silently hangs quite often and I can't seem to track why or when. or like when some script may fail but won't report a failure and so my save-dmesg script never runs but i can't see why :) | 20:47 |
sadmac2 | first one sounds like an fsck issue | 20:48 |
Zeqadious | probably | 20:49 |
Zeqadious | aha | 21:01 |
Zeqadious | status job should return something right? | 21:48 |
sadmac | I think so | 21:51 |
sadmac | yes. the manpage confirms | 21:51 |
sadmac | initctl status job | 21:51 |
sadmac | (there's no symlink for status, I don't think) | 21:51 |
Zeqadious | initctl status job returns nothing too | 21:52 |
Zeqadious | it should shouldn't it? | 21:52 |
sadmac | on 0.5 it might only if the job is running | 21:52 |
Zeqadious | returns nothing if the job is running | 21:53 |
Zeqadious | returns nothing if the job isn't running too | 21:53 |
sadmac | weeeeird | 21:53 |
sadmac | what does it return with no argument? | 21:54 |
Zeqadious | I think the problem this indicates is what is making it so difficult for me to write the scripts. RESULT=ok for any respawning service doesn't work | 21:54 |
Zeqadious | initctl status returns all jobs with either running or not running | 21:54 |
sadmac | ok, so that works | 21:54 |
sadmac | ah, so the respawn flag kills the RESULT variable | 21:55 |
sadmac | bug. | 21:55 |
Zeqadious | aha | 21:55 |
Zeqadious | also my scripts are all in directories inside jobs.d like jobs.d/service/hald so the job is service/hald | 21:55 |
Zeqadious | but because of the mangle bug status service/hald returns nothing | 21:56 |
Zeqadious | just found that status service_2fhald works though | 21:56 |
sadmac | yeah. that's not a new one. | 21:56 |
Zeqadious | yeah i saw that one in the bug list | 21:56 |
Zeqadious | this is just the first instance where i had to type the mangled form in to get it to work | 21:57 |
keesj | respawn is broken in many ways in 0.5 | 22:02 |
Zeqadious | well i filed a bug | 22:08 |
Zeqadious | 5pm works out. thanks for the help all, i'm going home. | 22:09 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!