On 02/15/2012 03:20 AM, Sage Weil wrote:
On Tue, 14 Feb 2012, Mandell Degerness wrote:
How big can the monitor data directory grow? Is there a rule of thumb
that can be used?
Right now the monitor dir include teh cluster log files (log and log.*),
which grow indefinitely unless you rotate them. The other directories
are mostly trimmed by the monitor itself and should stay pretty small (few
hundred megs at most?). The lingering exception is the mdsmap dir, which
isn't yet trimmed, but you guys won't be using that anyway (and the maps
are small).
Maybe Wido or someone else who has a long running cluster can see how big
it is?
I just took a look at my oldest cluster, which has been running for
about 4 months now with a single monitor.
This cluster has 6 OSD's and is running two virtual machines with RBD.
The monitor directory is 29M right now.
On my larger cluster, which I had to re-install, the monitor directories
are about 400M.
The logm directory is however trimmed? My oldest messages are from
February 14th on both clusters.
I'd however advice somewhere around 5GB at at least. I'm running my
monitors on ext4 and I've ran out of inodes a couple of times. But if
all directories are trimmed, you are not likely to run into that.
You should however be careful with the log.* files, they can grow
quickly. We could provide a standard logrotate script in the deb's /
rpm's for those?
Wido
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
