Page MenuHomePhabricator

Service diamond creates 500+ MByte /var/log/diamond/diamond.log
Closed, ResolvedPublic

Description

I had to delete those files and restart the diamond service on the Tools host because they were clogging up /var/log.

Chase explained in #wikimedia-operations:

<scfc_de> chasemp: What is /var/log/diamond/diamond.log and is it necessary?
<chasemp> scfc_de: it's the local log of the statistics polling, it's not
necessary in that it won't work without it, but it doesn't logging
those details to stdout seemingly so upstart logging isn't useful in
those cases. it's handy but not could-never-live-without

So I'd propose that on Labs, by default this log isn't generated. A Puppet variable could override this default if there is wider demand for that.


Version: unspecified
Severity: normal

Details

Reference
bz66458

Event Timeline

bzimport raised the priority of this task from to Needs Triage.Nov 22 2014, 3:17 AM
bzimport set Reference to bz66458.
bzimport added a subscriber: Unknown Object (MLST).

On a randomly chosen instance, I see:

-rw-r--r-- 1 diamond nogroup 9810916 Jul 2 23:59 archive.log.2014-07-02
-rw-r--r-- 1 diamond nogroup 9810168 Jul 3 23:59 archive.log.2014-07-03
-rw-r--r-- 1 diamond nogroup 9811659 Jul 4 23:59 archive.log.2014-07-04
-rw-r--r-- 1 diamond nogroup 9806164 Jul 5 23:59 archive.log.2014-07-05
-rw-r--r-- 1 diamond nogroup 9813206 Jul 6 23:59 archive.log.2014-07-06
-rw-r--r-- 1 diamond nogroup 16792013 Jul 7 14:06 diamond.log.2014-07-06
-rw-r--r-- 1 diamond nogroup 6444463 Jul 7 15:45 archive.log
-rw-r--r-- 1 diamond nogroup 1237334 Jul 8 16:46 diamond.log.2014-07-08
drwxr-xr-x 2 diamond root 4096 Jul 11 19:42 .
-rw-r--r-- 1 diamond nogroup 126 Jul 11 19:43 diamond.log

Since it's the 17th, I presume this has somehow been fixed?

Yes, Chase and Yuvi fiddled with diamond quite a bit with the conclusion (IIRC) that it doesn't log at all on Labs.

It does log, but only logs errors. We killed the archive handler that logged all the *metrics* being sent, which was causing the huge log files.