| ssarah | hei guys | 14:52 |
|---|---|---|
| ssarah | i want to ask, i wana use upstart instead of node forever. How can i make it that upstart only tries to restart a process a number of times before calling for support? | 14:53 |
| ssarah | anyone? =O | 14:58 |
| slangasek | ssarah: what do you mean by "calling for support"? | 17:06 |
| slangasek | you can use the "respawn limit" command to control how many times upstart will retry a failing process; but I'm not sure if that's what you're asking for | 17:07 |
| ssarah | slangasek: i mean, that after failing for several times, it does a custom command | 18:01 |
| ssarah | get it? :) | 18:02 |
| slangasek | ok | 18:02 |
| slangasek | upstart is certainly flexible enough that this should be possible, but I can't point to a ready-made example of someone doing this before | 18:02 |
| slangasek | in fact I think you probably would need to keep a counter, outside of upstart itself, unless the failures are immediate and could be done through 'respawn limit' | 18:03 |
| xnox | ssarah: i'd recommend use existing monitoring to monitor that service is up, when upstart gives up restarting it the status will change to "stop/waiting" and your monitoring system (nagios, check-mk, etc) can take appropriate action for automated recovery and/or further escalation. | 18:12 |
| xnox | ssarah: at least, in my previous sys-admin days that's what we did. Since nagios already does smart escalation notification and has appropriate means to encode shifts and rotas. | 18:12 |
| ssarah | xnox: what status changes to "stop/waiting" | 18:13 |
| ssarah | ? | 18:13 |
| xnox | ssarah: of your job. If you start it, it becomes "$ status your-job" will say start/running. And would stay like that, and respawned as needed. | 18:17 |
| xnox | ssarah: when it fails permamently, it will change status to "$ status your-job" will become "stop/waiting", and thus your nagios monitoring can be checking that. | 18:18 |
| ssarah | im sorry if this sounds too basic but that variable is something i can see where? | 18:19 |
| xnox | ssarah: type on the command line: $ sudo status tty2 | 18:30 |
| xnox | ssarah: $ sudo stop tty2 | 18:30 |
| xnox | ssarah: $ sudo status tty2 | 18:30 |
| xnox | ssarah: $ sudo start tty2 | 18:30 |
| xnox | ssarah: $ sudo status tty2 | 18:30 |
| xnox | ssarah: $ cat /etc/init/tty2.conf | 18:30 |
| xnox | ssarah: for more about jobs, their states, what states they transition through and how to control/inspect/observe that see http://upstart.ubuntu.com/cookbook/ | 18:31 |
| xnox | ssarah: for more control you can add your custom jobs, or e.g. monitor signals, events and notifications on the DBUS interface | 18:31 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!