| dkeith | http://www.amazon.com/Avoid-Huge-Ships-John-Trimmer/product-reviews/0870334336/ref=dp_top_cm_cr_acr_txt?ie=UTF8&showViewpoints=1 | 13:40 |
|---|---|---|
| dkeith | ignore this - my chat client's gone wild | 13:41 |
| qwebirc75393 | im going to install mythbuntu on my pc connected to my telly and connect it to my router, will windows see it from my laptop | 15:56 |
| slipcon | Just looking for a confirmation on an upgrade path... I'm running Mythbuntu 11.04, mythtv 0.24-fixes. I think I eventually want to end up on 12.04 LTS, but that only supports 0.25+... so would my path be to upgrade mythtv to 0.25 using dpkg-reconfigure mythbuntu-repos, and then once thats working well use 'do-release-upgrade' to go from 11.04 to 11.10 to 12.04? | 17:32 |
| mrand | slipcon: yes, you could do that if you want to disconnect the mythtv functionality upgrade from the OS upgrade. Or you could just do the release upgrades and it would also get you to 0.25. | 18:40 |
| dekarl | tgm4883: superm1, any opinion on adding http://paste.ubuntu.com/1659151/ to master packagings upstart script as a workaround to the backend not stopping for a long time after being killed? (I'm just preparing the packaging/script so I can test it this weekend) | 20:00 |
| tgm4883 | dekarl, I've looked at that today. I was planning on asking in the upstart channel if there was a way to ensure something dies when it's killed | 20:02 |
| tgm4883 | dekarl, or is that not to fix the issue with the backend not dieing when requested | 20:02 |
| dekarl | aye | 20:03 |
| dekarl | I think the issue might be that upstart only waits 5 seconds and the moves on | 20:03 |
| dekarl | s/the moves/then moves/ | 20:03 |
| dekarl | from http://upstart.ubuntu.com/cookbook/#stopping-a-job "Upstart waits for up to kill timeout seconds (default 5 seconds) for the process to end." | 20:04 |
| tgm4883 | dekarl, so extending that would fix it? | 20:05 |
| tgm4883 | dekarl, cause we can do that http://upstart.ubuntu.com/cookbook/#kill-timeout | 20:05 |
| tgm4883 | dekarl, I thought the problem is we ask it to stop and it doesn't stop and then upstart loses track of it | 20:06 |
| dekarl | is there a state transition diagram somewhere with all the parameters and states? I find it a hard to get it all together | 20:06 |
| tgm4883 | dekarl, how should we be asking the backend to stop? by default, it uses SIGTERM | 20:06 |
| dekarl | "kill signal" is the normal signal but "kill timeout" is how long upstart waits before killing... doesnt make sense to me | 20:07 |
| tgm4883 | dekarl, http://upstart.ubuntu.com/cookbook/#job-states | 20:07 |
| tgm4883 | dekarl, so you ask the backend to stop "via service mythtv-backend stop" it then waits 5 seconds and then sends SIGTERM | 20:08 |
| tgm4883 | dekarl, I'm assuming that timeout is to allow any scripts in the job to finish running | 20:08 |
| dekarl | sounds like 5 seconds for the pre-stop script | 20:08 |
| tgm4883 | yea | 20:08 |
| tgm4883 | so we can change it to send something else other than SIGTERM, or we can extend that timeout | 20:09 |
| tgm4883 | but IDK if that timeout thing is what is causing it | 20:09 |
| tgm4883 | maybe we should send something better than SIGTERM | 20:09 |
| dekarl | I think the post-stop script that sends a kill after some time of waiting isn't so bad. It gives the backend some time for the "really doing something" part but cuts off the "taking ages" part early | 20:11 |
| dekarl | one could argue that 4 seconds is a bit on the short side and change it to 30 or something | 20:12 |
| dekarl | but I haven't timed the shutdown with a stop watch, yet :) | 20:12 |
| dekarl | hmm, according to http://upstart.ubuntu.com/cookbook/#stopping-a-job SIGTERM is sent and after 5 seconds of waiting SIGKILL is sent. | 20:16 |
| tgm4883 | hmm | 20:18 |
| tgm4883 | superm1, ping | 20:18 |
| tgm4883 | dekarl, when the backend continues to run, is it a zombie process? | 20:19 |
| tgm4883 | dekarl, superm1 I'd imagine that when it sends SIGTERM and SIGKILL, it's sending it to the mythbackend process, not the mythbackend.real process | 20:19 |
| tgm4883 | I wonder if that is causing some of the problems | 20:20 |
| dekarl | dinner is ready, bbl | 20:23 |
| tgm4883 | I bet we need to do something special with http://upstart.ubuntu.com/cookbook/#expect | 20:23 |
| tgm4883 | dekarl, I've got something for you to test when you get back | 21:03 |
| dekarl | tgm4883: re | 21:12 |
| tgm4883 | tgm4883> rhpot1991, yea, just see what PID 'sudo service mythtv-backend status' reports | 21:12 |
| tgm4883 | <tgm4883> rhpot1991, and if that is the right PID for mythbackend | 21:12 |
| tgm4883 | dekarl, can you do the same ^^ | 21:12 |
| dekarl | mythtv-backend start/running, process 4322 | 21:13 |
| dekarl | 4323 ? Sl 30:14 /usr/bin/mythbackend --syslog local7 --user mythtv --verbose channel,eit,record,siparser | 21:13 |
| dekarl | 4322 ? Ss 0:00 /bin/sh -e /proc/self/fd/9 | 21:13 |
| tgm4883 | awesome | 21:13 |
| tgm4883 | dekarl, ok, that is what I expected | 21:13 |
| tgm4883 | and I have a fix for that | 21:13 |
| dekarl | expect fork? | 21:13 |
| tgm4883 | dekarl, no | 21:14 |
| tgm4883 | LANG=$LANG /usr/bin/mythbackend --syslog local7 --user mythtv | 21:14 |
| tgm4883 | that should be | 21:14 |
| tgm4883 | LANG=$LANG exec /usr/bin/mythbackend --syslog local7 --user mythtv | 21:14 |
| tgm4883 | dekarl, can you change to that, restart the backend service and check again | 21:14 |
| dekarl | I can test it in about 15 minutes | 21:15 |
| tgm4883 | dekarl, great thanks :) | 21:16 |
| tgm4883 | dekarl, I think that should fix the issue we're seeing. Upstart is sending the SIGTERM to the wrong process | 21:16 |
| tgm4883 | which still works most of the time | 21:16 |
| dekarl | mythtv-backend start/running, process 6883 | 21:28 |
| dekarl | 6883 ? Ssl 0:00 /usr/bin/mythbackend --syslog local7 --user mythtv --verbose channel,eit,record,siparser | 21:28 |
| dekarl | no it waits longer when stopping :) and there is no mythbackend once its done waiting | 21:29 |
| tgm4883 | dekarl, sweet | 21:29 |
| dekarl | s/no it/now it/ | 21:29 |
| tgm4883 | dekarl, previously you were able to reproduce the issue with the backend not stopping when asked? | 21:29 |
| dekarl | appears to wait 5 seconds and then kill... just as documented :) | 21:29 |
| tgm4883 | \o/ | 21:30 |
| dekarl | the service command would return basically immidiatly from stopping | 21:30 |
| dekarl | and the backend would hang around for a minute or two | 21:30 |
| tgm4883 | dekarl, good to know. So this is much much better :) | 21:31 |
| dekarl | long enough for dpkg to perform an update of all new debs from the packaging process and start the new version | 21:31 |
| tgm4883 | I'm going to see if I can get that fix pushed fast | 21:31 |
| dekarl | http://paste.ubuntu.com/1659781/ I can push it if you like | 21:32 |
| tgm4883 | dekarl, you've got access (forgive me, I'm not sure who all the mythtv developers are) | 21:32 |
| tgm4883 | err, that was a question ^ | 21:32 |
| dekarl | aye, I can push to packaging | 21:33 |
| tgm4883 | is 5 seconds enough to wait? | 21:33 |
| dekarl | But I'm happy to cut carrots for the baby girl instead :) | 21:33 |
| dekarl | that's something I can't answer quickly, might ask over in #mythtv | 21:33 |
| tgm4883 | will do | 21:33 |
| dekarl | Ok, I'll cut carrots then :) | 21:34 |
| tgm4883 | dekarl, thanks, I might bug you to push that later :) | 21:34 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!