For example:
/var/log/refinery/sqoop-mediawiki-private.log
For example:
/var/log/refinery/sqoop-mediawiki-private.log
Subject | Repo | Branch | Lines +/- | |
---|---|---|---|---|
Rotate logs in refinery based on time rather than size | operations/puppet | production | +4 -2 |
We do have a logrotate config on an1003:
elukey@analytics1003:/etc/logrotate.d$ cat refinery # Note: This file is managed by Puppet. # /var/log/refinery/*.log { size 100M rotate 4 missingok notifempty nocreate su hdfs analytics-admins }
It logrotates on data size and not on date (like daily, hourly, etc..). So if we want to change it, the file is in: profile/files/analytics/refinery-logrotate.conf
Change 464732 had a related patch set uploaded (by Nuria; owner: Nuria):
[operations/puppet@production] Rotate logs in refinery based on time rather than size
@elukey : I think (please let me know if you disagree) that a logrotate on time is more intuitive than size so we have a clear point in time when rotation will happen for all logs. Set it to rotate logs weekly and keep 10 logs so there would be 2 months of logs at any one time.
Change 464732 merged by Elukey:
[operations/puppet@production] Rotate logs in refinery based on time rather than size