To be sure to understand, if I create 2 times replicated pool toto with 1024 pgs and 1 pgp, pg and data of pool toto will be mapped on only 2 OSDs and on 2 servers right ? Sent from my iPhone > On 21 mai 2015, at 18:58, Florent MONTHEL <fmonthel@xxxxxxxxxxxxx> wrote: > > Thanks Ilya for this clear explanation! > I'm searching that for a long time > > Best practices is to have pg = pgp in order to "avoid" using of the same set of osd right ? (On a small cluster you will have) > > Sent from my iPhone > >>> On 21 mai 2015, at 07:49, Ilya Dryomov <idryomov@xxxxxxxxx> wrote: >>> >>> On Thu, May 21, 2015 at 12:12 PM, baijiaruo@xxxxxxx <baijiaruo@xxxxxxx> wrote: >>> Re: what's the difference between pg and pgp? >> >> pg-num is the number of PGs, pgp-num is the number of PGs that will be >> considered for placement, i.e. it's the pgp-num value that is used by >> CRUSH, not pg-num. For example, consider pg-num = 1024 and pgp-num >> = 1. In that case you will see 1024 PGs but all of those PGs will map >> to the same set of OSDs. >> >> When you increase pg-num you are splitting PGs, when you increase >> pgp-num you are moving them, i.e. changing sets of OSDs they map to. >> >> Thanks, >> >> Ilya >> _______________________________________________ >> 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
 |