On Wed, Nov 27, 2013 at 5:56 AM, Alphe Salas <asalas@xxxxxxxxx> wrote: > Hello all, > bad news the problem shown again today ! > I had mounted a bindfs and nfs server running on the same "proxy" server > when I ran a massive chmod on a directory with large amount of data I got > part of the directories that disapeared! > > The problem wasn t showing as fast as it was with kernel 3.11 but > it still was showing. > > I don t know the origine and if the bindfs / nfs are related. > How may entries in the directories are missing. When this issue happens, please cd the cephfs mount point and check if there are missing entries. Reagrds Yan, Zheng > > Alphe Salas > I.T ingeneer > > > On 11/22/13 10:15, Alphe Salas Michels wrote: >> >> Hello Yan, >> Good guess ! thank you for your advice I updated this morning my >> cephfs-proxy ubuntu 13.10 to recommended 3.12 kernel and >> after the first preliminary tests the issue isn t showing anymore. >> >> >> Regards, >> signature >> >> *Alphé Salas* >> Ingeniero T.I >> >> >> *<http://www.kepler.cl>* >> >> On 11/21/13 23:06, Yan, Zheng wrote: >>> >>> On Fri, Nov 22, 2013 at 9:19 AM, Alphe Salas Michels <asalas@xxxxxxxxx >>> <mailto:asalas@xxxxxxxxx>> wrote: >>> >>> Hello all! >>> >>> I experience a strange issue since last update to ubuntu 13.10 >>> (saucy) and ceph emperor 0.72.1 >>> >>> kernel version 3.11.0-13-generic #20-Ubuntu >>> >>> ceph packages installed are the ones for RARING >>> >>> when I mount my ceph cluster using cephfs and I upload a tons of >>> data or do a directory listing (find . -printf "%d %k" ) or do >>> a chown -R user:user * at some point the filesystem disapear! >>> >>> I don t know how to solve this issue there is no entry in anylog >>> the only thing that seems to be affected is ceph-watch-notice that >>> get stuckl >>> and forbid the unmount "have to pid kill .-9 that process to >>> umount / mount the ceph cluster on client proxy to start over the >>> process. >>> in the chown if I put --changes to slow it down just enought then >>> the problem seems to disapear. >>> >>> >>> sounds like the d_prune_aliases() bug. please try updating 3.12 kernel >>> or using ceph-fuse >>> >>> Yan, Zheng >>> >>> Any suggestions are welcome >>> Atte, >>> >>> -- >>> *Alphé Salas* >>> Ingeniero T.I >>> >>> Descripción: cid:image001.gif@01CAA59C.F14CE4B0*Kepler Data Recovery* >>> >>> *Asturias 97, Las Condes** >>> Santiago- Chile** >>> **(56 2) 2362 7504* >>> >>> asalas@xxxxxxxxx <mailto:asalas@xxxxxxxxx> >>> *www.kepler.cl <http://www.kepler.cl>* >>> >>> >>> _______________________________________________ >>> ceph-users mailing list >>> ceph-users@xxxxxxxxxxxxxx <mailto: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
 |