Hello. I have a relatively new ceph installation that is only running ceph FS at the moment. We are seeing intermittent issues where "ceph -s" is reporting "MDS report slow requests" and sometimes the MDS crash and take a while to recover/replay or we have to manually restart an mds service to get the state back to HEALTH_OK. Is there any documentation for recommended configuration? Here is our cluster setup: 35 total nodes, 88 cores, 512GB ram, 100Gb network 2 ceph fs data pools, 1 is all ssd, the other is nvme 3 active MDS, 1 pinned to the nvme pool/dir, 1 pinned to another large directory, and the third has no pinning 2 standby MDS ceph config dump: mds advanced mds_beacon_grace 60.000000 mds basic mds_cache_memory_limit 68719476736 mds advanced mds_cache_trim_threshold 65536 mds advanced mds_recall_max_decay_rate 2.000000 Please let me know if more info is required. Thanks! _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |