| === ion_ is now known as ion | ||
| === JanC_ is now known as JanC | ||
| djszapi | how about Keybuk ? :) | 09:09 |
|---|---|---|
| === apw` is now known as apw | ||
| djszapi | where does upstart not print the process ID in few cases ? | 11:49 |
| djszapi | s/where/why/ | 11:50 |
| djszapi | process id with exec lines and no process id with script lines ? | 11:53 |
| ion | It always tells the pid it’s tracking. | 12:28 |
| djszapi | what do you mean ion ? | 12:29 |
| djszapi | sysconfig-middle start/running | 12:33 |
| djszapi | gconfd start/running, process 438 | 12:33 |
| djszapi | I have got lines like this, one of them with no PID. | 12:33 |
| ion | sysconfig-middle probably has no main process. | 12:44 |
| djszapi | how can that be ? | 12:45 |
| djszapi | some main process should be in any case, at least I thought that way :o | 12:46 |
| ion | Nope, it’s optional. | 12:46 |
| djszapi | I thought a tool is run from script section or with exec. | 12:47 |
| djszapi | and there is a main process in both cases. | 12:47 |
| ion | yes | 12:47 |
| djszapi | I am confused then :-S | 12:47 |
| djszapi | What other way is there ? | 12:48 |
| ion | script/exec is optional. | 12:48 |
| djszapi | is, but both have main process. | 12:48 |
| djszapi | My another issue is that when I try to enable the "--verbose" boot command-line option then I do not get this kind of messages (we discussed on Friday): Jan 14 14:52:49 deathspank init: tty6 main process (8557). | 12:52 |
| ion | sysconfig-middle doesn’t seem to have a main process, judging from the fact that what you pasted from initctl status/list says so. | 12:54 |
| ion | Try initctl log-priority [PRIORITY] | 12:55 |
| djszapi | well, it should do the same like verbose. | 12:55 |
| djszapi | "initctl log-priority warn" | 12:55 |
| djszapi | but tried it out, did not help. | 13:00 |
| djszapi | I definitely think what was run from a script section that does not have PID as said in the beginning. | 13:02 |
| djszapi | but that is why they should have in warn, verbose mode a 'Jan 14 14:52:49 deathspank init: tty6 main process (8557)'-like entry which I do not have... | 13:06 |
| djszapi | so it should show the PID of the shell, ttyS6 oslt. | 13:11 |
| djszapi | for example: sysconfig-middle has a parent process and I would like to know that's PID. | 13:22 |
| djszapi | * obviously has | 13:23 |
| djszapi | any idea anybody, please ? | 14:41 |
| robs_ | hi | 14:50 |
| robs_ | I have a problem with upstart "stop" and "start" commands inside a 10.04 Vserver Guest | 14:51 |
| robs_ | it seems it's the same problem outlined here | 14:51 |
| robs_ | http://www.freak-search.com/de/thread/2643371/vserver_problems_installing_ubuntu_10.10_as_a_guest_of_a_debian_host | 14:51 |
| robs_ | I have the same problem (ie stop "service" or start "service" hangs), this is the debug I get from "stop dbus" inside a ubuntu 10.04 guest: http://pastebin.com/AjA2sAKG | 14:52 |
| JanC | robs_: issues with upstart inside a vserver are a known problem | 15:06 |
| robs_ | JanC: following http://linux-vserver.org/Upstart_issues the machine boots and halts just fine | 15:07 |
| robs_ | JanC: only "start" and "stop" hang | 15:07 |
| djszapi | I cannot believe I need to use audit or other kernellevel stuff directly to get the PID of the parent for a command that was run from the script field of upstart. | 15:10 |
| djszapi | it just seems a bit khm.. :) | 15:10 |
| djszapi | * upstart configuration file. | 15:10 |
| djszapi | no idea why it cannot be the part of upstart :( | 15:11 |
| JanC | djszapi: are you sure the main script is still running? | 15:11 |
| djszapi | how does that matter ? | 15:12 |
| djszapi | syslog entry should be done when it runs ? | 15:12 |
| JanC | and upstart is not 1.0 yet, some features are still on the TODO list ;) | 15:12 |
| djszapi | I know but I have been told on Friday it is possible with verbose cli boot parameter | 15:12 |
| djszapi | and it seems that statement 1) was not true 2) I do somehing wrong | 15:13 |
| djszapi | JanC: do you have any idea ? | 15:17 |
| djszapi | Keybuk mentioned me on Friday: with verbose I should get like: Jan 14 14:52:49 deathspank init: tty6 main process (8557) in syslog | 15:18 |
| djszapi | where tty6 is the 'sh' where the command was run from. | 15:18 |
| djszapi | from the upstart config file's script entry. | 15:18 |
| djszapi | btw, where is it the best to ask questions if this channel is not OK for that ? | 15:39 |
| JanC | djszapi: all the jobs I see with start/running and no PID are jobs that have no main exec/script, but only pre/post-start/stop | 15:40 |
| JanC | and this channel is fine for asking questions, alternatively you can try the mailing list | 15:40 |
| djszapi | well I did not still get any answer. | 15:41 |
| djszapi | from this morning. | 15:41 |
| djszapi | the question is still the same, how can I get the process id of the parent ? | 15:42 |
| djszapi | wherever it was run from, let it be pre/post-start/stop | 15:42 |
| JanC | are you sure it is still running? | 15:43 |
| djszapi | I need the pid when it was running. | 15:44 |
| djszapi | aegis-loader:191,sh:188,init:1 -> that is the call tree | 15:45 |
| djszapi | and I would like to get 188 back. | 15:45 |
| djszapi | aegis start/running | 15:45 |
| djszapi | aegis.conf run the aegis-loader in a post-start like way | 15:45 |
| djszapi | but I need to know the PID of the parent ('sh') | 15:46 |
| djszapi | do you understand ? | 15:46 |
| JanC | yes ☺ | 15:46 |
| djszapi | define "yes ?" :) | 15:47 |
| djszapi | otherwise I would say 'yes'. | 15:47 |
| JanC | what you apparently see as a "?" was a smiley | 15:48 |
| djszapi | pardon ? | 15:48 |
| JanC | yes, I understand what you want now | 15:48 |
| djszapi | do you know the answer how to do it ? | 15:49 |
| djszapi | verbose is supposed to do that. | 15:49 |
| djszapi | also initctl log-priority warn | 15:49 |
| djszapi | but none of them works :( | 15:49 |
| djszapi | JanC: feel free to help me, if you can :P | 15:59 |
| JanC | djszapi: I did some test, and for me it lists pre-start shell PIDs ? | 16:09 |
| JanC | djszapi: maybe you don't need 'warn' but 'info'? | 16:10 |
| JanC | warn should only show warnings & errors? | 16:12 |
| djszapi | JanC: verbose should work anyway. | 16:13 |
| djszapi | but I am trying with info | 16:13 |
| JanC | 'debug' will show even more, but 'info' should be enough | 16:14 |
| djszapi | does it work with 'warn' by you ? | 16:15 |
| djszapi | no pid with info either. | 16:16 |
| djszapi | JanC: http://pastebin.com/HVGiQHqz | 16:17 |
| djszapi | which is supposed to be the PID of the 'sh' line ? | 16:18 |
| djszapi | captrace lines are internal stuff, you can skip them, not that important tho. | 16:19 |
| JanC | http://pastebin.com/4CSGKYhm --> you can see the PID (13885) of the shell that executes the post-stop script (which only runs "sleep 10", hence it takes exactly 10 seconds to run) | 16:28 |
| JanC | that's with log-priority info | 16:30 |
| JanC | didn't test 'warn', but from the docs 'warn' shouldn't show this ;) | 16:31 |
| JanC | "`warn' (ordinary messages are suppressed whilst still logging warnings and errors);" | 16:31 |
| JanC | does your dbus job have any pre/post-stop scripts? | 16:32 |
| djszapi | re | 17:33 |
| djszapi | JanC: I do not really understand why you keep repeating this pre/post-start/stop thingy | 17:33 |
| djszapi | it should return the PID of the parent in *any* case | 17:33 |
| JanC | djszapi: it seems like it currently doesn't log the PID of the shell indeed | 18:12 |
| JanC | in that case | 18:12 |
| djszapi | you mean in my pastebin, right ? | 18:13 |
| JanC | not only in your pastebin, also on my system | 18:13 |
| djszapi | that is a BUG | 18:14 |
| djszapi | Scott mentioned it should. | 18:14 |
| djszapi | JanC: what to do then ? | 18:49 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!