Hi, I've upgraded our test cluster to Octopus, and enabled the auto-scaler. It's nearly finished: PG autoscaler decreasing pool 11 PGs from 1024 to 32 (4d) [==========================..] (remaining: 3h) But I notice it looks to be making pool 11 smaller when HEALTH_WARN thinks it should be larger: root@sto-t1-1:~# ceph health detail HEALTH_WARN 1 pools have many more objects per pg than average; 9 pgs not deep-scrubbed in time [WRN] MANY_OBJECTS_PER_PG: 1 pools have many more objects per pg than average pool default.rgw.buckets.data objects per pg (313153) is more than 23.4063 times cluster average (13379) ...which seems like the wrong thing for the auto-scaler to be doing. Is this a known problem? Regards, Matthew More details: ceph df: root@sto-t1-1:~# ceph df --- RAW STORAGE --- CLASS SIZE AVAIL USED RAW USED %RAW USED hdd 993 TiB 782 TiB 210 TiB 211 TiB 21.22 TOTAL 993 TiB 782 TiB 210 TiB 211 TiB 21.22 --- POOLS --- POOL ID STORED OBJECTS USED %USED MAX AVAIL .rgw.root 2 69 KiB 4 1.4 MiB 0 220 TiB default.rgw.control 3 1.1 MiB 8 3.3 MiB 0 220 TiB default.rgw.data.root 4 115 KiB 14 3.6 MiB 0 220 TiB default.rgw.gc 5 5.3 MiB 32 23 MiB 0 220 TiB default.rgw.log 6 31 MiB 184 96 MiB 0 220 TiB default.rgw.users.uid 7 249 KiB 8 1.8 MiB 0 220 TiB default.rgw.buckets.data 11 23 GiB 10.02M 2.0 TiB 0.30 220 TiB rgwtls 13 54 KiB 3 843 KiB 0 220 TiB pilot-metrics 14 285 MiB 2.60M 476 GiB 0.07 220 TiB pilot-images 15 40 GiB 4.97k 122 GiB 0.02 220 TiB pilot-volumes 16 192 GiB 48.90k 577 GiB 0.09 220 TiB pilot-vms 17 125 GiB 33.79k 376 GiB 0.06 220 TiB default.rgw.users.keys 18 111 KiB 5 1.5 MiB 0 220 TiB default.rgw.buckets.index 19 4.0 GiB 246 12 GiB 0 220 TiB rbd 20 39 TiB 10.09M 116 TiB 14.88 220 TiB default.rgw.buckets.non-ec 21 344 KiB 1 1.0 MiB 0 220 TiB rgw-ec 22 7.0 TiB 1.93M 11 TiB 1.57 441 TiB rbd-ec 23 45 TiB 11.73M 67 TiB 9.22 441 TiB default.rgw.users.email 24 23 MiB 1 69 MiB 0 220 TiB pilot-backups 25 73 MiB 3 219 MiB 0 220 TiB device_health_metrics 26 51 MiB 186 153 MiB 0 220 TiB root@sto-t1-1:~# ceph osd pool autoscale-status POOL SIZE TARGET SIZE RATE RAW CAPACITY RATIO TARGET RATIO EFFECTIVE RATIO BIAS PG_NUM NEW PG_NUM AUTOSCALE .rgw.root 70843 3.0 992.7T 0.0000 1.0 32 on default.rgw.control 1116k 3.0 992.7T 0.0000 1.0 32 on default.rgw.data.root 115.1k 3.0 992.7T 0.0000 1.0 32 on default.rgw.gc 5379k 3.0 992.7T 0.0000 1.0 32 on default.rgw.log 32036k 3.0 992.7T 0.0000 1.0 32 on default.rgw.users.uid 248.7k 3.0 992.7T 0.0000 1.0 32 on default.rgw.buckets.data 23894M 3.0 992.7T 0.0001 1.0 32 on rgwtls 55760 3.0 992.7T 0.0000 1.0 32 on pilot-metrics 285.3M 3.0 992.7T 0.0000 1.0 32 on pilot-images 41471M 3.0 992.7T 0.0001 1.0 32 on pilot-volumes 192.3G 3.0 992.7T 0.0006 1.0 32 on pilot-vms 124.6G 3.0 992.7T 0.0004 1.0 32 on default.rgw.users.keys 111.1k 3.0 992.7T 0.0000 1.0 32 on default.rgw.buckets.index 4090M 3.0 992.7T 0.0000 1.0 32 on rbd 39430G 3.0 992.7T 0.1164 1.0 1024 on default.rgw.buckets.non-ec 344.3k 3.0 992.7T 0.0000 1.0 32 on rgw-ec 7175G 1.5 992.7T 0.0106 1.0 64 on rbd-ec 45806G 1.5 992.7T 0.0676 1.0 1024 on default.rgw.users.email 23530k 3.0 992.7T 0.0000 1.0 32 on pilot-backups 74699k 3.0 992.7T 0.0000 1.0 32 on device_health_metrics 52128k 3.0 992.7T 0.0000 1.0 32 on -- The Wellcome Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.