Hello, It seems that cyrus over non-linux NFS may have better chance. But NFS on Linux got some improvements after 2.6.15. NFSv4 for kernel 3.2 got significant lock improvements as listed below. So, keeping metadata over local fs, and only metadata partition over NFS may get better chance of success. But it is based on assumption that metadata partition needs the file locking more than data partition. Is it correct? Good explanations of NFS file locking problems on Linux . http://unix.stackexchange.com/questions/1777/flock2-versus-fcntl2-over-a-nfs http://0pointer.de/blog/projects/locking.html http://0pointer.de/blog/projects/locking2 http://www.samba.org/samba/news/articles/low_point/tale_two_stds_os2.html syscall stress tester http://codemonkey.org.uk/projects/trinity/ http://nfs.sourceforge.net/ http://linux-nfs.org http://wiki.linux-nfs.org/wiki/index.php/CITI_Experience_with_Directory_Delegations nfsd improvements in linux kernel 3.2 http://www.spinics.net/lists/linux-nfs/msg24895.html linux mandatory locking mount option ? http://en.wikipedia.org/wiki/File_locking http://en.wikipedia.org/wiki/File_locking#Problems old linux kernel locking explanation https://bugzilla.redhat.com/show_bug.cgi?id=59992 a cyrus over NFS on non linux (OpenIndiana? Solaris? seems to use zones) http://lists.andrew.cmu.edu/pipermail/info-cyrus/2012-June/036242.html a cyrus over NFS on linux https://lists.andrew.cmu.edu/pipermail/info-cyrus/2012-March/035951.html Regards. Andre Felipe Machado http://www.techforce.com.br
---- Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/ To Unsubscribe: https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus
