| codebeaker | hi all, I'm getting into upstart for the first time, and I wonder if I've missed something… I'm trying to use upstart to keep monit running | 10:42 |
|---|---|---|
| codebeaker | hi all, I'm getting into upstart for the first time, and I wonder if I've missed something… I'm trying to use upstart to keep monit running (re-post, since I was D/C'ed) | 12:02 |
| SpamapS | codebeaker: what do you think you're missing? | 12:46 |
| codebeaker | if I kill monit, upstart doesn't bring it back | 12:46 |
| codebeaker | which I gathered was the point of upstart | 12:46 |
| codebeaker | ( in addition to actually starting it in the firstplace) | 12:46 |
| SpamapS | did you put the word 'respawn' in your job file? | 12:46 |
| SpamapS | no respawning isn't the only point of upstart.. its mostly just to be feature compatible w/ sysvinit | 12:47 |
| SpamapS | Its just a little smarter about it than sysvinit. :) | 12:47 |
| codebeaker | ok, I'll check - also what is the /correct/ place for upstart configs? | 12:47 |
| codebeaker | on my ubuntu 10.4 system they appear to live in /etc/init/ | 12:47 |
| codebeaker | but I've seen lots of talk online about using /etc/event.d/ | 12:47 |
| SpamapS | they must be in /etc/init since Ubuntu 9.10 | 12:48 |
| codebeaker | ok, good to know - thanks :) | 12:48 |
| codebeaker | *bow* | 12:48 |
| SpamapS | You may want to submit your monit job as a feature-request bug against the package if it works out for you. | 12:49 |
| codebeaker | not sure, it's 3 lines of code :) | 12:50 |
| codebeaker | (well, four now :)) | 12:50 |
| codebeaker | SpamapS: what's your personal opinion on Monit's place in a modern hosting env | 12:53 |
| codebeaker | I'm basically using monit to keep my services up, because I know it well enough | 12:53 |
| SpamapS | Its quite suited for that, especially since it is capable of deeper monitoring than upstart's "is the process alive" | 12:55 |
| codebeaker | ok, perfect… so upstart to keep the system running, and monit to keep things working at the application layer | 13:02 |
| SpamapS | Yeah | 13:04 |
| wraiden | does anyone know if upstart 1.0 ist compatible wrt utmp handling when i use agetty? | 13:12 |
| wraiden | the teminal sessions aren't decremented in issue output anymore since i upgraded to 1.0 | 13:12 |
| wraiden | i'll try to switch to mingetty today to check it isn't an incompatibility with util-linux's agetty as mingetty was mentioned in some changelog... | 13:15 |
| SpamapS | doesn't login handle utmp maintenance? | 13:15 |
| wraiden | looks like it doesn't if i read the changes that went in from plautrbach correctly... | 13:17 |
| wraiden | mybe i have a wrong tty.conf in /etc/init | 13:18 |
| SpamapS | ahh I'm looking now.. ok.. utmp is indeed touched by init now | 13:18 |
| SpamapS | I say now like I know it didn't before.. maybe it did before.. :p | 13:18 |
| wraiden | the problem was that issue incremented the open tty session counters but doesnt decremented it after the loginsession was closed and upstart respawned the getty ... | 13:19 |
| wraiden | this bug was in upstart from the start i think | 13:20 |
| wraiden | don't know how sysv handled it | 13:20 |
| plautrba | wraiden: can you paste somewhere 'utmpdump /var/run/utmp' ? | 13:20 |
| SpamapS | right looks like now if a job process terminates upstart looks in utmp and fixes it | 13:20 |
| plautrba | wraiden: afaik it should work regardless of used *getty | 13:20 |
| wraiden | hello :) i'll open a session to the box an paste the things... | 13:21 |
| wraiden | http://pastebin.com/LfDLA8hR | 13:24 |
| wraiden | cant look at the issue output right now as i'm not in front of the box | 13:25 |
| plautrba | this looks ok, what is concretely your problem? | 13:33 |
| wraiden | if i login an a local tty the issue displays the logged in users ok but if i logout the logged in users stay the same and doesn't get decremented | 13:35 |
| wraiden | this behavor war fixed in the first round of patches that you attached to the bugreport ... | 13:36 |
| wraiden | i tested them and used them as patches against 0.6 but with the upgrade to version 1.0 i went back to the buggy behavior | 13:37 |
| wraiden | i'll post another utmp dunp when i have access to the box again around 18:30 CET | 13:38 |
| plautrba | in 1.0 is almost same patch | 13:39 |
| plautrba | and utmp table says that there is 2 correctly restarted local tty session, 1 closed remote, and 1 active remote session | 13:40 |
| wraiden | thats right | 13:41 |
| plautrba | command 'w' should show one logged user | 13:41 |
| wraiden | i rebooted the box | 13:41 |
| wraiden | kernel upgrade... | 13:41 |
| wraiden | the wrong behavior was befor the reboot | 13:41 |
| wraiden | i'll repro the wron output and post an updated dump | 13:42 |
| wraiden | should i mail it or crate a bug report? | 13:42 |
| plautrba | mail to mailling list or bug report are fine, or you can first try to ping me here | 13:43 |
| wraiden | k, will look here first | 13:43 |
| wraiden | anyway, thanks for your work about utmp handling, this one bugged me a long time :) | 13:44 |
| plautrba | i've just setup my fedora rawhide box to use aggety and it works ok | 13:44 |
| wraiden | exec /sbin/agetty -I '\033(K' tty1 38400 | 13:44 |
| wraiden | is what i use | 13:44 |
| wraiden | maybe it's related to that | 13:44 |
| plautrba | I use only /sbin/agetty /dev/tty6 38400 but -I should not affects that | 13:46 |
| wraiden | i'll repro the problem tonight and post the details along the dump... | 13:48 |
| atpa8a | hello! | 22:34 |
| atpa8a | i'm binding slapd to an ip alias and it seems that aliases come up late on boot which causes slapd fail to start... can i start networking "syncronously" or something? | 22:34 |
| Keybuk | no, but you can have slapd start after networking | 22:44 |
| atpa8a | i think that's how it's configured... | 22:47 |
| atpa8a | or should i create /etc/init/slapd.conf for that? | 22:47 |
| Keybuk | you would need an /etc/init/slapd.conf or something in Upstart that starts slapd | 22:48 |
| atpa8a | i don't have that right now | 22:48 |
| atpa8a | just learning upstart | 22:48 |
| atpa8a | thought it was related to insserv | 22:49 |
| atpa8a | hmm | 22:55 |
| atpa8a | so slapd is not an upstart job right now | 22:56 |
| atpa8a | which means it's handled from /etc/init/rc.conf, right? | 22:56 |
| mbiebl | no, /etc/init.d/slapd | 22:56 |
| atpa8a | true :) i misspoke | 22:57 |
| mbiebl | atpa8a: only indirectly via /etc/init/rc.conf | 22:57 |
| atpa8a | right | 22:57 |
| atpa8a | that's what i meant | 22:57 |
| atpa8a | can i just add something like start on (net-device-up...) to /etc/init/rc.conf to make sure the networking is up before anything else? | 22:58 |
| atpa8a | or change /etc/init/rc-sysinit.conf from 'start on filesystem and net-device-up IFACE=lo' to 'start on filesystem and net-device-up' | 23:01 |
| atpa8a | basically i want *all* of the network up before slapd starts | 23:03 |
| atpa8a | that didn't help | 23:14 |
| atpa8a | any hints? | 23:15 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!