---------- Forwarded message ---------- From: Brad Hubbard <bhubbard@xxxxxxxxxx> Date: Fri, Jun 1, 2018 at 9:24 PM Subject: Re: inconsistent pgs :- stat mismatch in whiteouts To: shrey chauhan <shrey.chauhan@xxxxxxxxxxxxx> Cc: ceph-users <ceph-users@xxxxxxxx> Too late for me today. If you send your reply to the list someone else may provide an answer more expansive than the following. An object flagged as a "whiteout" logically does not exist. I believe this is a way for a cache tier to cache the fact that this object does not exist so it does not have to forward IO requests for that object to the backing cache, it can just immediately notify the client that the object does not exist. Not sure why the scrub is uncovering mismatching values for whiteouts, but that's what is happening. On Fri, Jun 1, 2018 at 8:58 PM, shrey chauhan <shrey.chauhan@xxxxxxxxxxxxx> wrote: > yes it is pool. > > > moreover what are these whiteouts? and when does this mismatch happen? > > > > On Fri, Jun 1, 2018 at 3:51 PM, Brad Hubbard <bhubbard@xxxxxxxxxx> wrote: >> >> On Fri, Jun 1, 2018 at 6:41 PM, shrey chauhan >> <shrey.chauhan@xxxxxxxxxxxxx> wrote: >> > Hi, >> > >> > I keep getting inconsistent placement groups and every time its the >> > whiteout. >> > >> > >> > cluster [ERR] 9.f repair stat mismatch, got 1563/1563 objects, 0/0 >> > clones, >> > 1551/1551 dirty, 78/78 omap, 0/0 pinned, 12/12 hit_set_archive, 0/-9 >> > whiteouts, 28802382/28802382 bytes, 16107/16107 hit_set_archive bytes. >> > >> > I tried looking it up and I hardly found anything on this. >> > >> > What are these whiteouts and when do they start causing inconsistent >> > issues? >> >> Seems to be related to cache tiering. Is pool 9 a cache pool or does >> it have a relation to a cache pool (or has it in the past)? >> >> > >> > Thanks >> > >> > _______________________________________________ >> > ceph-users mailing list >> > ceph-users@xxxxxxxxxxxxxx >> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >> > >> >> >> >> -- >> Cheers, >> Brad > > -- Cheers, Brad -- Cheers, Brad _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com