Re: filestore split settings

[David Turner <drakonstein@xxxxxxxxx>]

More important than being able to push those settings or further is probably the ability to actually split your subfolders. I've been using variants of this [1] script I created a while back to take care of that.

To answer your question, we do run with much larger settings than you're using. 128/-16. The negative prevents subfolder merging while still allowing the value to be used for calculations the splitting number.

Take a look at the script. It stops the osds, aggressively sets the subfolder settings, splits the subfolders to that setting, puts your settings back, and starts your osds. I do this about once a month for our use case of growing data.

[1] https://gist.github.com/drakonstein/cb76c7696e65522ab0e699b7ea1ab1c4

On Wed, Aug 22, 2018, 7:37 AM Rafael Lopez <rafael.lopez@xxxxxxxxxx> wrote:

Hi all,

For those still using filestore and running clusters with a large number of objects, I am seeking some thoughts on increasing the filestore split settings. Currently we have:

filestore merge threshold = 70

filestore split multiple = 20

Has anyone gone higher than this?

We are hitting the threshold of 22400 files per dir on osds for a particular pool and experiencing slow reqs, and other osd badness as a result. I am wondering if we can simply increase these values to push the files per dir threshold and delay splitting without major consequences, eg. to 80/30. This would probably buy us enough time to move to bluestore.

--

Rafael Lopez

Research Devops Engineer

Monash University eResearch Centre

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com