Ceph/ODF: MDS が OOM またはメモリー不足、あるいは "1 clients failing to respond to capability release" 状態までメモリーを確保する

Issue

Ceph MDS は、MDS をホストしているノードで、OOM が発生するか、メモリー枯渇が発生するまで、非常に高速にメモリーを確保します。

同じコードの問題は、"1 clients failing to respond to capability release" という "HEALTH_WARN" として現れることがあります。

-bash 5.1 $ ceph -s
  cluster:
    id:     8d23xxxx-redacted-cluster-ID-yyyya00794f2
    health: HEALTH_WARN
            1 clients failing to respond to capability release

  services:
    mon: 3 daemons, quorum cefesp000003,cefesp000002,cefesp000001 (age 11w)
    mgr: cefesp000003(active, since 2M), standbys: cefesp000002, cefesp000001
    mds: 1/1 daemons up, 2 standby
    osd: 321 osds: 321 up (since 12d), 321 in (since 2w)

  data:
    volumes: 1/1 healthy
    pools:   14 pools, 13729 pgs
    objects: 106.82M objects, 61 TiB
    usage:   189 TiB used, 421 TiB / 610 TiB avail
    pgs:     13728 active+clean
             1     active+clean+scrubbing+deep

  io:
    client:   729 MiB/s rd, 345 MiB/s wr, 2.89k op/s rd, 4.47k op/s wr

-bash 5.1 $ ceph health detail
HEALTH_WARN 1 clients failing to respond to capability release
[WRN] MDS_CLIENT_LATE_RELEASE: 1 clients failing to respond to capability release
    mds.cephfs.cefesp000002.gruurg(mds.0): Client smeesp000032.mydomain.org:csi-cephfs-node failing to respond to capability release client_id: 390442015