On Mon, Apr 15, 2013 at 12:19 PM, Amit Vijairania <amit.vijairania@xxxxxxxxx> wrote: > Hello! > > How is pg_num (pgp_num) for default pools (data, metadata and rbd) is > calculated? > > According to following: > http://ceph.com/docs/master/rados/operations/placement-groups > > If my cluster has 30 OSDs and 3 replica per Object (osd pool default size = > 3), I should be creating new pools with pg_num = pgp_num = 1000.. But, > default pools on my cluster are: In my opinion, this formula is just a suggestion, which means ~100 PGs per OSD may be better. > $ ceph osd pool get data pg_num > pg_num: 1984 > $ ceph osd pool get metadata pg_num > pg_num: 1984 > $ ceph osd pool get rbd pg_num > pg_num: 1984 > > Is there a different formula used for default pools? pg_num of a pool can increase by pg split, so 1984 here shouldn' t be the original value. I'm also not clear about "best practice" about pg_num and pg splitting, hoping for more info. > Thanks! > Amit > > > _______________________________________________ > 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
 |