Clean up memcached supervisor logrotate

[Paul Everitt]

From discussion with gocept:

No, KARL does not get restarted. SIGUSR2 does the following for supervisord. From http://supervisord.org/running.html#signal-handlers:

"...supervisord will close and reopen the main activity log and all child log files...."

I don’t think I need a postrotate to kill my supervisord (and thus memcached) during log rotation. I plan to install:

/srv/osfkarl/memcachedconfs/var/log/*.log {
daily
missingok
rotate 10
compress
delaycompress
notifempty
copytruncate
}

I think SIGUSR2 is a good idea in this case, too. At least I would test if your memcached the really gets killed when its parent (supervisord) gets a SIGUSR2. I suspect that this will work, just like it does for the KARL application. You can trigger the rotation by executing "user-logrotate -v" as the osfkarl/karlstaging service user (forcing the rotation works by adding -f to the command). A PID comparision (before-after) should help finding out whether memcached is retarted or not.

Apart from that, the snippets you define in /var/spool/logrotate/<service-user> also have underlying default options (see /etc/logrotate.options) which you do not need to redefine in your script.

[Paul Everitt (paul-agendaless)]

This is actually working well enough.