The pid files are kind-of immobile. The right place for them under RedHat based distributions would be under /var/run but even if I specify '-p /var/run/glusterfs/glusterd.pid' when starting glusterd, that has no effect on the locations of the nfs, glusterfsd, glusterfs or gsync pid files. I feel the better structure is to have them based on a pid-directory structure rather than volume/service based. For instance: {pid-root}/glusterd.pid {pid-root}/nfs.pid (or {pid-root}/nfs/nfs.pid) {pid-root}/gsync/volume1-server2-data-volume1.pid {pid-root}/bricks/server1-data-brick1.pid {pid-root}/bricks/server1-data-brick2.pid {pid-root}/fuse/volume1-rb_dst_brick.pid The command-line option I would envision to go along with that would then be '--pid-root'.
priority will be re-addressed after 3.3.0 GA
I have run into a related problem. The gluster-swift packages by kkeithley create a 'swift.pid' file in a subdirectory of /var/lib/glusterd/vols/$volname. When you issue "gluster volume delete $volname" it is unable to delete the volume directory because of this file. I have filed bug 861497 for the general problem with the volume directory not getting fully removed on volume deletion. General comment: It seems like a very bad idea for one package (gluster-swift) to put its pid file in a directory that is created and managed at runtime by another package (gluster).
because of the large number of bugs filed against mainline version\ is ambiguous and about to be removed as a choice. If you believe this is still a bug, please change the status back to NEW and choose the appropriate, applicable version for it.
*** This bug has been marked as a duplicate of bug 1258561 ***