On Aug 14, 2012, Sage Weil <s...@inktank.com> wrote: > Do you mind trying to reproduce this on the current master?
Tried, but couldn't. It looks like it is indeed fixed in master, but not in 0.50. > BTW, if you have a script or anything you were using to test, that'd be > great to add to our test suite. It occurred occasionally while rsyncing (with -aH) trees containing hardlinks into the ceph filesystem, but it became a real problem when I decided to stop using snapshots and use hardlink farms instead. Creating a hardlink farm of a tree containing hardlinks (like /usr/share/zoneinfo) presumably compounds the problem, but I've hit it most often after creating linkfarms of a few backups of my phone's root directory, some old SHR GNU/Linux release, probably just because those have enough directories and files to get stuff kicked out of the mds cache. You could probably trigger it with something like this: rsync -aH /usr/share/zoneinfo/ /mnt/ceph/sub1/ rsync -aH /usr/share/zoneinfo/ /mnt/ceph/sub2/ --link-dest=/mnt/ceph/sub1/ umount /mnt/ceph # ceph-fuse survives the umount, so wait # till it flushes everything to the mds: while killall -0 ceph-fuse; do sleep 1; done restart mds # maybe not needed mount /mnt/ceph # kick all of sub1/ and sub2/ out of the cache somehow, and then: ls -lR /mnt/ceph/sub2 # freezes # restarting mds at this point may get it stuck at rejoin -- Alexandre Oliva, freedom fighter http://FSFLA.org/~lxoliva/ You must be the change you wish to see in the world. -- Gandhi Be Free! -- http://FSFLA.org/ FSF Latin America board member Free Software Evangelist Red Hat Brazil Compiler Engineer -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
