Ceph » bluestore

Delete all pools (containing 40 million of objects in total) and wait for the recycling of used space to complete. Each OSD boot takes more than half an hour because it is in the process of clear_ temp_ object.

Each OSD rocksdb has used about 5GB of space. After manual compaction, the startup time returns to normal.

Only the PG of EC pool takes a long time:

2022-05-06T15:41:21.487+0800 7f647bafea40 20 osd.7 1480  clearing temps in 17.fs2_head pgid 17.fs2
2022-05-06T15:41:21.487+0800 7f647bafea40 15 bluestore(/var/lib/ceph/osd/ceph-7) collection_list 17.fs2_head start GHMIN end GHMAX max 64
2022-05-06T15:41:21.487+0800 7f647bafea40 20 bluestore(/var/lib/ceph/osd/ceph-7) _collection_list range 2#-19:f0000000::::0#0 to 2#-19:f1000000::::0#0 and 2#17:f0000000::::0#0 to 2#17:f1000000::::0#0 start GHMIN
2022-05-06T15:42:12.679+0800 7f647bafea40 20 bluestore(/var/lib/ceph/osd/ceph-7) _collection_list pend 2#-19:f1000000::::0#0
2022-05-06T15:42:12.679+0800 7f647bafea40 20 bluestore(/var/lib/ceph/osd/ceph-7) _collection_list oid 2#17:18000000::::head# >= 2#-19:f1000000::::0#0
2022-05-06T15:42:12.679+0800 7f647bafea40 20 bluestore(/var/lib/ceph/osd/ceph-7) _collection_list oid 2#17:f0000000::::head# end GHMAX
2022-05-06T15:42:13.126+0800 7f647bafea40 20 bluestore(/var/lib/ceph/osd/ceph-7) _collection_list oid 2#17:fd000000::::head# >= 2#17:f1000000::::0#0
2022-05-06T15:42:13.126+0800 7f647bafea40 10 bluestore(/var/lib/ceph/osd/ceph-7) collection_list 17.fs2_head start GHMAX end GHMAX max 64 = 0, ls.size() = 1, next = GHMAX