On Tue, Jun 23, 2015 at 12:17 PM, Lionel Bouton <lionel-subscription@xxxxxxxxxxx> wrote: > On 06/23/15 11:43, Gregory Farnum wrote: >> On Tue, Jun 23, 2015 at 9:50 AM, Erik Logtenberg <erik@xxxxxxxxxxxxx> wrote: >>> Thanks! >>> >>> Just so I understand correctly, the btrfs snapshots are mainly useful if >>> the journals are on the same disk as the osd, right? Is it indeed safe >>> to turn them off if the journals are on a separate ssd? >> That's not quite it...it *is* safe to turn off btrfs snapshots, but by >> doing so you get the same behavior as XFS does by default. > > I just disabled snapshots and the OSD logged this: > > mount: enabling WRITEAHEAD journal mode: checkpoint is not enabled > > which I assume means that I don't have to change the following > configuration parameters, the OSD takes care of using sensible values > for them: > > filestore journal parallel > filestore journal writeahead Write. You probably shouldn't mess with these in any case; the OSD selects the right mode based on other things. > > From the limited feedback I got from our monitoring our disk writes are > now ~1MB/s instead of ~4MB/s when the cluster is mostly idle. There are > still spikes of activity (compared to XFS) but they might just be linked > to the default btrfs commit delay and are harmless. Xfs OSDs still have > a lower amount of writes though but this is expected when comparing a > COW filesystem to a classic one. > > Note that these numbers might push you from Intel DC S3500 to S3610 (for > example) if you plan to use btrfs on Intel SSD OSDs: ~1MB/s is 30+TB/year... > With btrfs snapshots enabled and 4MB/s this is 120+TB/year. > > Lionel _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |