Hey ceph-users, may I ask (nag) again about this issue?  I am wondering if anybody can confirm my observations? I raised a bug https://tracker.ceph.com/issues/54136, but apart from the assignment to a dev a while ago here was not response yet. Maybe I am just holding it wrong, please someone enlighten me. Thank you and with kind regards Christian On 02/02/2022 20:10, Christian Rohmann wrote:

Hey ceph-users,

I am debugging a mgr pg_autoscaler WARN which states a target_size_bytes on a pool would overcommit the available storage. There is only one pool with value for  target_size_bytes (=5T) defined and that apparently would consume more than the available storage:

--- cut --- # ceph health detail HEALTH_WARN 1 subtrees have overcommitted pool target_size_bytes [WRN] POOL_TARGET_SIZE_BYTES_OVERCOMMITTED: 1 subtrees have overcommitted pool target_size_bytes     Pools ['backups', 'images', 'device_health_metrics', '.rgw.root', 'redacted.rgw.control', 'redacted.rgw.meta', 'redacted.rgw.log', 'redacted.rgw.otp', 'redacted.rgw.buckets.index', 'redacted.rgw.buckets.data', 'redacted.rgw.buckets.non-ec'] overcommit available storage by 1.011x due to target_size_bytes 15.0T on pools ['redacted.rgw.buckets.data']. --- cut ---

But then looking at the actual usage it seems strange that 15T (5T * 3 replicas) should not fit onto the remaining 122 TiB AVAIL:

--- cut --- # ceph df detail --- RAW STORAGE --- CLASS  SIZE     AVAIL    USED     RAW USED  %RAW USED hdd    293 TiB  122 TiB  171 TiB   171 TiB      58.44 TOTAL  293 TiB  122 TiB  171 TiB   171 TiB      58.44

--- POOLS --- POOL                             ID  PGS   STORED   (DATA) (OMAP)   OBJECTS  USED     (DATA)   (OMAP)   %USED  MAX AVAIL QUOTA OBJECTS  QUOTA BYTES  DIRTY  USED COMPR  UNDER COMPR backups                           1  1024   92 TiB   92 TiB  3.8 MiB   28.11M  156 TiB  156 TiB   11 MiB  64.77     28 TiB N/A            N/A            N/A      39 TiB      123 TiB images                            2    64  1.7 TiB  1.7 TiB  249 KiB  471.72k  5.2 TiB  5.2 TiB  748 KiB   5.81     28 TiB N/A            N/A            N/A         0 B          0 B device_health_metrics            19     1   82 MiB      0 B   82 MiB       43  245 MiB      0 B  245 MiB      0     28 TiB N/A            N/A            N/A         0 B          0 B .rgw.root                        21    32   23 KiB   23 KiB 0 B       25  4.1 MiB  4.1 MiB      0 B      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.control             22    32      0 B      0 B 0 B        8      0 B      0 B      0 B      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.meta                23    32  1.7 MiB  394 KiB  1.3 MiB    1.38k  237 MiB  233 MiB  3.9 MiB      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.log                 24    32   53 MiB  500 KiB   53 MiB    7.60k  204 MiB   47 MiB  158 MiB      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.otp                 25    32  5.2 KiB      0 B  5.2 KiB        0   16 KiB      0 B   16 KiB      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.buckets.index       26    32  1.2 GiB      0 B  1.2 GiB    7.46k  3.5 GiB      0 B  3.5 GiB      0     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.buckets.data        27   128  3.1 TiB  3.1 TiB 0 B    3.53M  9.5 TiB  9.5 TiB      0 B  10.11     28 TiB N/A            N/A            N/A         0 B          0 B redacted.rgw.buckets.non-ec      28    32      0 B      0 B 0 B        0      0 B      0 B      0 B      0     28 TiB N/A            N/A            N/A         0 B          0 B --- cut ---

I then looked at how those values are determined at https://github.com/ceph/ceph/blob/9f723519257eca039126a20aa6a2a7d2dbfb5dba/s.... Apparently "total_bytes" are compared with the capacity of the root_map. I added a debug line and found that the total in my cluster was already at:

  total=325511007759696

so in excess of 300 TiB - Looking at "ceph df" again this usage seems strange.

Looking at how this total is calculated at https://github.com/ceph/ceph/blob/9f723519257eca039126a20aa6a2a7d2dbfb5dba/s..., you see that the larger value (max) of "actual_raw_used" vs. "target_bytes*raw_used_rate" is determined and then summed up.

I dumped the values for all pools my cluster with yet another line of debug code:

---cut --- pool_id 1 - actual_raw_used=303160109187420.0, target_bytes=0 raw_used_rate=3.0 pool_id 2 - actual_raw_used=5714098884702.0, target_bytes=0 raw_used_rate=3.0 pool_id 19 - actual_raw_used=256550760.0, target_bytes=0 raw_used_rate=3.0 pool_id 21 - actual_raw_used=71433.0, target_bytes=0 raw_used_rate=3.0 pool_id 22 - actual_raw_used=0.0, target_bytes=0 raw_used_rate=3.0 pool_id 23 - actual_raw_used=5262798.0, target_bytes=0 raw_used_rate=3.0 pool_id 24 - actual_raw_used=162299940.0, target_bytes=0 raw_used_rate=3.0 pool_id 25 - actual_raw_used=16083.0, target_bytes=0 raw_used_rate=3.0 pool_id 26 - actual_raw_used=3728679936.0, target_bytes=0 raw_used_rate=3.0 pool_id 27 - actual_raw_used=10035209699328.0, target_bytes=5497558138880 raw_used_rate=3.0 pool_id 28 - actual_raw_used=0.0, target_bytes=0 raw_used_rate=3.0 --- cut ---

All values but those of pool_id 1 (backups) make sense. For backups it's just reporting a MUCH larger actual_raw_used value than what is shown via ceph df. The only difference of that pool compared to the others is the enabled compression:

--- cut --- # ceph osd pool get backups compression_mode compression_mode: aggressive --- cut ---

Apparently there already was a similar issue (https://tracker.ceph.com/issues/41567) with a resulting commit (https://github.com/ceph/ceph/commit/dd6e752826bc762095be4d276e3c1b8d31293eb0)

changing which from "bytes_used" to the "stored" field for "pool_logical_used".

But how does that take compressed (away) data into account? Does "bytes_used" count all the "stored" bytes, summing up all uncompressed bytes for pools with compression? This surely must be a bug then, as those bytes are not really "actual_raw_used".

I was about to raise a bug, but I wanted to ask here on the ML first if I misunderstood the mechanisms at play here. Thanks and with kind regards,

Christian