2024-07-30T13:52:08.814Z

<Utkarsh Bhatt> Hey everyone, I have a test Ceph lab with uneven OSDs (not of same size). I am getting a strange issue when trying to add a new pool with `'ceph', '--id', 'admin', 'osd', 'pool', 'create', '--pg-num-min=32', 'gnocchi', '64'.`
```Error ERANGE: pg_num 64 size 3 for this pool would result in 272 cumulative PGs per OSD (1635 total PG replicas on 6 'in' root OSDs by crush rule) which exceeds the mon_max_pg_per_osd value of 250```

2024-07-30T13:52:37.962Z

<Utkarsh Bhatt> ```$ sudo ceph osd df
ID  CLASS  WEIGHT   REWEIGHT  SIZE     RAW USE  DATA     OMAP  META     AVAIL    %USE  VAR   PGS  STATUS
 1         0.27280   1.00000  279 GiB  215 MiB  146 MiB   0 B   70 MiB  279 GiB  0.08  1.12  108      up
 6         0.36389   1.00000  373 GiB  264 MiB  226 MiB   0 B   38 MiB  372 GiB  0.07  1.03  158      up
 5         0.90970   1.00000  932 GiB  592 MiB  544 MiB   0 B   48 MiB  931 GiB  0.06  0.92  322      up
 4         0.43669   1.00000  447 GiB  302 MiB  264 MiB   0 B   38 MiB  447 GiB  0.07  0.98  171      up
 3         0.21829   1.00000  224 GiB  230 MiB  178 MiB   0 B   52 MiB  223 GiB  0.10  1.49   93      up
 2         0.87129   1.00000  892 GiB  564 MiB  520 MiB   0 B   44 MiB  892 GiB  0.06  0.92  303      up
                       TOTAL  3.1 TiB  2.1 GiB  1.8 GiB   0 B  290 MiB  3.1 TiB  0.07 ```

2024-07-30T13:53:20.724Z

<Utkarsh Bhatt> I am not quite sure what's happening, but I doubt that the large variance in OSD disk size may be the cause ?

2024-07-30T15:46:57.443Z

<yuriw> `reef` builds failing <https://shaman.ceph.com/builds/ceph/wip-yuri8-testing-2024-07-30-0629-reef/>

2024-07-30T15:47:16.049Z

<yuriw> `reef` builds failing <https://shaman.ceph.com/builds/ceph/wip-yuri8-testing-2024-07-30-0629-reef/>
anybody else is seeing this?

2024-07-30T15:48:29.401Z

<Casey Bodley> similar complaint in <#C1HFJ4VTN|>

2024-07-30T19:11:42.896Z

<Æmerson> I'm not going to do it now, but long term, since the new versions of opentelemetry-cpp no longer support the Jaeger exporter, does anyone have any idea whether we'd be better off with OTLP HTTP, OTLP gRPC, or both and let the user configure it?

2024-07-30T20:42:57.006Z

<Casey Bodley> cc @Deepika Upadhyay @Yuval Lifshitz