Description of problem: MDS leaks file descriptors across exec which causes it to run out after several respawns. Version-Release number of selected component (if applicable): 3.0 How reproducible: 100%. Respawn mds a few dozen times via `ceph mds fail 0`. Use single MDS cluster (no standby) to see more easily. Steps to Reproduce: 1. while sleep 0.5; do ceph mds fail 0; done 2. MDS will eventually fail to create an event file descriptor as noted in the log and then quit. Actual results: Exits with failure. Expected results: MDS continues respawning infinitely.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:3530