Ceph » RADOS

My monitor keeps growing. After it exceeds 300GB, it cannot be started because the disk is not enough. After the space is restored, two (mon-t, mon-u) of the original three monitor (mon-r, mon-t, mon-u) services cannot work normally, and it has been in the electing stage.

Then I select mon-t data and perform the following operations:

ceph-mon -i t --mon-data /var/lib/ceph/mon/ceph-t/data --extract-monmap /tmp/monmap

monmaptool /tmp/monmap --rm r
monmaptool /tmp/monmap --rm u

ceph-mon -i t --mon-data /var/lib/ceph/mon/ceph-t/data --inject-monmap /tmp/monmap

Finally, restart mon-t, but the store.db directory is still growing, and it has exceeded 400GB in 5 hours. During the entire time, in addition to the short up/down of some osd, ceph also accompanied a large number of warnings about slow ops on mon-t, and all pgs were in active+clean and scrubbing state.

And neither the offline ceph-monstore-tool nor the online compact command can make the space occupied by mon-t smaller

Ceph version: v14.2.11