Hi Adam, Could you add "debug mds = 20" in your ceph.conf and retry? There should also be a file "core" in /, this is a dump which you can use to make a backtrace with gdb (if you have the debug symbols). gdb /usr/bin/cmds /core gdb> bt Could you try this? -- Met vriendelijke groet, Wido den Hollander CTO Contact: http://www.pcextreme.nl/contact On Fri, 2010-07-09 at 19:22 +0000, Adam C. Emerson wrote: > I know it's not supported, but I tried it out to see what kind of > error behaviour I would get. > > The following were preformed on a simple ceph filesystem (one monitor, > one mds, one osd all on the same machine) running current unstable > with the filesystem mounted via cfuse: > > /mnt# mkdir foo > /mnt# cd foo > /mnt/foo# cat > something > These are the contents of the file. > /mnt/foo# cd .snap > /mnt/foo/.snap# mkdir old > /mnt/foo/.snap# cd .. > /mnt/foo# cat > something > These are different contents of the file. > /mnt/foo# cd .snap > /mnt/foo/.snap# ls > old/ > /mnt/foo/.snap# cd old > root@seasalt:/mnt/foo/.snap/old# ls > something > /mnt/foo/.snap/old# cat something > These are the contents of the file. > /mnt/foo/.snap/old# cat > something > zsh: invalid argument: something > /mnt/foo/.snap/old# cat > somthingelse > These are divergent contents. > > This last command hangs, and in the MDS log I find: > mds/CDir.cc: In function 'CDentry* CDir::add_null_dentry(const std::string&, snapid_t, snapid_t)': > mds/CDir.cc:225: FAILED assert(lookup_exact_snap(dname, last) == 0) > > Thank you very much. > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
