| === erkules_ is now known as erkules | ||
| jodh | /topic Upstart 1.9 | http://upstart.ubuntu.com/cookbook/ | Post to mailing list if no response here: http://bit.ly/ooeNVv | 13:45 |
|---|---|---|
| jodh | 13:45 | |
| === jodh changed the topic of #upstart to: Upstart 1.9 | http://upstart.ubuntu.com/cookbook/ | Post to mailing list if no response here: http://bit.ly/ooeNVv | ||
| elmo | jodh: \o/ | 15:19 |
| elmo | jodh: could I safely run upstart 1.9 on raring if I backport it? | 15:19 |
| xnox | elmo: wait for us to merge 1.9 into saucy & apply ubuntu-specific patches for apparmor, and then backport that package. | 15:22 |
| elmo | xnox: ok | 15:22 |
| xnox | elmo: well, i expect you will just be able to recompile for raring. | 15:23 |
| erkules | ahoi short question. having in his old init script an own option. Does upstart has any way to have own options? I was thinking about using emit, but that is too fuzzy. | 15:26 |
| xnox | erkules: what do you want to achieve? and what do you want to do? | 15:31 |
| xnox | erkules: all common tricks are described here: http://upstart.ubuntu.com/cookbook/#cookbook-and-best-practises | 15:32 |
| erkules | xnox: hi, I was reading a blogpost. There they patched galera-init with a bootstrap option. So when you start the script with that option it creates a new GTID | 15:32 |
| erkules | xnox: okidoki | 15:34 |
| xnox | erkules: i'm still confused what your initial problem/question is. =) | 15:36 |
| xnox | erkules: if you want to launch different instances of the same daemon you can be using instances | 15:37 |
| erkules | hehe | 15:37 |
| xnox | erkules: what is galera-init? what is GTID? | 15:37 |
| erkules | ok GTID (global transaction id) | 15:37 |
| erkules | it is a galera thing | 15:37 |
| erkules | mysql-multi-master | 15:38 |
| erkules | service mysql start -> starts the server with old GTID | 15:38 |
| erkules | service mysql start --bootstrap forces the server to create a new GTID | 15:39 |
| xnox | erkules: if a job is already running, upstart will not try starting it again. Thus you should stop; start. | 15:52 |
| xnox | erkules: if you want to keep the "old one" running. and run a new one with a different GTID simultaniously. Look into passing variables to jobs and/or instances. | 15:53 |
| xnox | erkules: it looks like above is an init.d script, are you converting it to a native upstart job? | 15:53 |
| xnox | erkules: cause under upstart, you can continue using init.d scripts and above should just work as upstream uses it. | 15:53 |
| erkules | xnox: I would like to implement the idea of extra options in my own upstart script | 16:04 |
| xnox | erkules: they are not scripts, and are not executed. | 16:07 |
| xnox | erkules: they are configuration files. | 16:08 |
| xnox | erkules: you can pass environment variables to them. | 16:08 |
| xnox | erkules: e.g. you can have $ start myjob BOOTSTRAP=true | 16:08 |
| xnox | but not arbitrary arguments. | 16:09 |
| * xnox slowly builds on disk. | 16:44 | |
| xnox | jodh: all 18 tests pass for me with saucy sbuild. | 16:48 |
| xnox | jodh: please note that for me locally upstart testsuite fails if sbuild is using: overlayfs, tmpfs, eatmydata. | 16:48 |
| erkules | xnox: thats cool | 16:48 |
| erkules | xnox++ | 16:49 |
| xnox | jodh: thus to build upstart I always use sbuild with a configuration closer to the distro builders: | 16:49 |
| xnox | jodh: so my /etc/schroot/chroot.d/sbuild-saucy-amd64 looks like this: http://paste.ubuntu.com/5808343/ | 16:50 |
| xnox | jodh: and I build upstart with sbuild -c saucy-plain-amd64 path-to.dsc | 16:50 |
| slangasek | xnox: so I confirm your result that the branch builds fine in my sbuild environment; must be something specific to eatmydata or other optimization | 20:02 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!