| p3rsist | If I want a service to be started after home directory is decrypted on my development machine and I also want to use that same init upstart conf for the server where there no encrypted home directory, does the filesystem event will work for both? | 00:34 |
|---|---|---|
| p3rsist | ion? | 00:50 |
| Jordan_U | Is there any way that https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/1077579 could be a bug in something other than upstart? Not reaping a child seems like a pretty basic failure of PID 1 so I'm surprised this hasn't (seemingly) recieved much attention. | 03:13 |
| xnox | Jordan_U: if pid 1 / upstart receives SIGCHLD for that process, it wil reap it. | 06:55 |
| xnox | Jordan_U: so something else has probably already received that signal and ignored it. | 06:56 |
| Jordan_U | xnox: So something like the process dying before becoming a child of PID 1? I would expect even then that PID 1 would reap the child, so I think I'm misunderstanding something. | 08:12 |
| xnox | Jordan_U: when should it reap it? upon a defunct process getting re-parented to pid1, i guess sigchild will never arrive in that case. | 08:17 |
| === marcoceppi_ is now known as marcoceppi | ||
| Jordan_U | xnox: I think a check for defunct processes when they're inherited would probably make sense. | 18:16 |
| kaimast | stupid question. how do i build upstart? | 22:30 |
| kaimast | simply running automake doesn't seem to work | 22:31 |
| xnox | kaimast: it's covered in the cookbook. I think you want $ autoreconf -f -i; ./configure --with-options-you-need --exec-prefix=you-need; make; make check. | 22:33 |
| xnox | kaimast: do run the extensive test-suite, that validates the build in many different ways. | 22:33 |
| kaimast | thanks xnox! :) | 22:33 |
| kaimast | do you also know if there are tests for extras? especially upstart-socket-bridge? | 22:34 |
| xnox | kaimast: http://upstart.ubuntu.com/cookbook/#development-and-testing should be helpful. | 22:34 |
| xnox | kaimast: i'm not sure check the code, the bridge itself are well just emitting events. Sockets are unit-tested in the libnih side of things, and fd passing in the events (as done by socket bridge) is unit tested in the init/ tests portion. | 22:35 |
| xnox | kaimast: (or actually possibly in the nih-dbus part of codebase) | 22:35 |
| xnox | kaimast: ideally one would do "imperical" integration tests for a bridge, the way it's done for re-exec in ./scripts/test/*.py which is in python unit tests =) | 22:37 |
| kaimast | wth? ./configure: line 2510: syntax error near unexpected token `[Copyright' | 22:52 |
| xnox | kaimast: do you have all dependencies installed? sudo apt-get build-dep upstart | 22:55 |
| xnox | kaimast: and did your "autoreconf -f -i" resulted without any severe warnings? | 22:56 |
| kaimast | yeah, btw in the cookbook it says "apt-get install build-dep" | 22:59 |
| xnox | kaimast: oops that's wrong =) | 23:00 |
| kaimast | okay rerun autoreconf and now it seems to do something :) | 23:05 |
| === fredbarns_ is now known as fredbarns | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!