| === robbiew is now known as robbiew_ | ||
| frog-o | dose upstart have anything to do with module loading | 19:04 |
|---|---|---|
| wasabi__ | Question: I have a job that needs to stop by running a command and waiting for the process to exit, but never sending TERM to the process. | 21:59 |
| wasabi__ | I think expect stop is the closest, except I'm not using SIGSTOP | 22:00 |
| wasabi__ | Oops | 22:01 |
| ion | Add a pre-stop script that runs the command and waits for the process to exit. | 22:01 |
| wasabi__ | Ahh. I can wait. Duh. | 22:02 |
| wasabi__ | Cool. So. Here's a harder one. I have a set of services which I need to attempt to shutdown, but if they fail to shutdown, I need the system back up. Can we do this yet? :0 | 22:19 |
| wasabi__ | (VirtualBox VMs) | 22:19 |
| wasabi__ | (they could fail to respond to acpipoweroff | 22:19 |
| Keybuk | wasabi__: yes, in the pre-stop | 22:43 |
| Keybuk | if failed_to_shutdown | 22:43 |
| Keybuk | then | 22:43 |
| wasabi__ | cool | 22:43 |
| Keybuk | start | 22:44 |
| Keybuk | fi | 22:44 |
| Keybuk | ;) | 22:44 |
| wasabi__ | does env support "" around values? | 22:47 |
| Keybuk | yes | 22:48 |
| wasabi__ | Hmm. Getting close. | 22:53 |
| wasabi__ | Got the job to start... but there's probably some unknown of fork | 22:53 |
| wasabi__ | And if 'start' is invoked, and the job actually does succeed to shutting down later, does it come back? I suppose respawn effects that? | 23:00 |
| wasabi__ | vm-station-1 main process (28695) killed by TERM signal | 23:03 |
| wasabi__ | grrrr | 23:03 |
| wasabi__ | oh i see | 23:03 |
| wasabi__ | pre-stop failed | 23:03 |
| wasabi__ | if pre-stop fails, shoudl stop continue? | 23:03 |
| Keybuk | no | 23:05 |
| Keybuk | pre-stop fails doesn't affect anything | 23:05 |
| wasabi__ | hmmm. | 23:16 |
| wasabi__ | main process terminated with status 5. weird. it's just exec sudo, followed by a command that does not fork. | 23:16 |
| wasabi__ | running it as root works. | 23:16 |
| wasabi__ | Keybuk, when you getting into Dallas? | 23:17 |
| wasabi__ | oh i know. sudo isn't staing alive. | 23:17 |
| wasabi__ | hmmm | 23:22 |
| wasabi__ | Variables are expanded on exec stanza's, I assume | 23:28 |
| wasabi__ | Hmm. So sudo is forking the task off, but it's getting the console, so you can't tell | 23:31 |
| wasabi__ | Hmm. Nope. | 23:35 |
| wasabi__ | sudo is execing | 23:36 |
| wasabi__ | could i have broken upstart? I have a job that just blocks on 'start' | 23:37 |
| wasabi__ | ISI\jhaltom@station-1:/etc/init$ sudo status vm-station-2 | 23:47 |
| wasabi__ | vm-station-2 stop/killed, process 9401 | 23:47 |
| wasabi__ | no 9401 exists | 23:47 |
| Keybuk | wasabi__: sunday evening | 23:54 |
| wasabi__ | so tomorrow | 23:55 |
| ion | http://heh.fi/tmp/workaround-upstart-snafu | 23:55 |
| ion | ./workaround-upstart-snafu 9401 | 23:55 |
| ion | :-P | 23:55 |
| wasabi__ | so it's a known issue? | 23:55 |
| ion | Upstart doesn’t really handle forking children very well at the moment. Your job is doing something else than what Upstart expects. The 0.10 release will fix that. | 23:57 |
| Keybuk | today, I mostly hate compiz | 23:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!