Symptom: Ceph is reporting a HEALTH_WARN and the warning message is 1 pools have many more objects per pg than average.
Cause:
When an additional pool to the store the CSM artifacts was introduced, it caused the amount of PGs created per OSD to not allow for the amount of data stored in that device. This will typically be seen on three node Ceph clusters since the number of PGs is set by the number of OSDs.
Fix:
Get the current pg_num and pgp_num value for the pool that is warning.
ceph osd pool get csm_admin_pool pg_num
ceph osd pool get csm_admin_pool pgp_num
Example Output:
pg_num: 32
pgp_num: 32
Set the new values.
ceph osd pool set csm_admin_pool pg_num 64
ceph osd pool set csm_admin_pool pgp_num 128
ceph osd pool set csm_admin_pool pg_num_min 64
ceph osd pool set csm_admin_pool pg_num_max 128
Expected Output:
NOTE: Each line in the below output is related to the line above and is only reflecting the output for each line above when it is executed.
set pool 15 pg_num to 64
set pool 15 pgp_num to 128
set pool 15 pg_num_min to 64
set pool 15 pg_num_max to 128