On Wed, Jan 29, 2020 at 7:33 AM Stefan Kooman <stefan@bit.nl> wrote:

Quoting Stefan Kooman (stefan@bit.nl):

...

Hi,

The command "ceph daemon mds.$mds perf dump" does not give the collection with MDS specific data anymore. In Mimic I get the following MDS specific collections:

- mds - mds_cache - mds_log - mds_mem - mds_server - mds_sessions

But those are not available in Nautilus anymore (14.2.4). Also not listed in a "perf schema".

Where did these metrics go?

Nobody? It would a shame to lose these helpfull metrics when upgrading to Nautilus (as none except prometheus metric collectors get (all) useful metrics and you need to reside to "local" daemon metrics for the time being ...).

wfm: # ceph daemon mds.`hostname -s` version {"version":"14.2.6","release":"nautilus","release_type":"stable"} # ceph daemon mds.`hostname -s` perf dump | jq 'keys[]' "AsyncMessenger::Worker-0" "AsyncMessenger::Worker-1" "AsyncMessenger::Worker-2" "cct" "finisher-PurgeQueue" "mds" "mds_cache" "mds_log" "mds_mem" "mds_server" "mds_sessions" "mempool" "objecter" "purge_queue" "throttle-msgr_dispatch_throttler-mds" "throttle-objecter_bytes" "throttle-objecter_ops" "throttle-write_buf_throttle" "throttle-write_buf_throttle-0x5575a16c5dc0" Maybe you're checking a standby MDS ? -- Dan