Thanks Casey. This helped me understand the purpose of this pool. I trimmed the usage logs which reduced the number of keys stored in that index significantly and I may even disable the usage log entirely as I don't believe we use it for anything.
On Fri, May 24, 2019 at 3:51 PM Casey Bodley <cbod...@redhat.com> wrote: > > > On 5/24/19 1:15 PM, shubjero wrote: > > Thanks for chiming in Konstantin! > > > > Wouldn't setting this value to 0 disable the sharding? > > > > Reference: http://docs.ceph.com/docs/mimic/radosgw/config-ref/ > > > > rgw override bucket index max shards > > Description:Represents the number of shards for the bucket index > > object, a value of zero indicates there is no sharding. It is not > > recommended to set a value too large (e.g. thousand) as it increases > > the cost for bucket listing. This variable should be set in the client > > or global sections so that it is automatically applied to > > radosgw-admin commands. > > Type:Integer > > Default:0 > > > > rgw dynamic resharding is enabled: > > ceph daemon mon.controller1 config show | grep rgw_dynamic_resharding > > "rgw_dynamic_resharding": "true", > > > > I'd like to know more about the purpose of our .usage pool and the > > 'usage_log_pool' in general as I cant find much about this component > > of ceph. > > You can find docs for the usage log at > http://docs.ceph.com/docs/master/radosgw/admin/#usage > > Unless trimmed, the usage log will continue to grow. If you aren't using > it, I'd recommend turning it off and trimming it all. > > > > > On Thu, May 23, 2019 at 11:24 PM Konstantin Shalygin <k0...@k0ste.ru> wrote: > >> in the config. > >> ``` "rgw_override_bucket_index_max_shards": "8",```. Should this be > >> increased? > >> > >> Should be decreased to default `0`, I think. > >> > >> Modern Ceph releases resolve large omaps automatically via bucket dynamic > >> resharding: > >> > >> ``` > >> > >> { > >> "option": { > >> "name": "rgw_dynamic_resharding", > >> "type": "bool", > >> "level": "basic", > >> "desc": "Enable dynamic resharding", > >> "long_desc": "If true, RGW will dynamicall increase the number of > >> shards in buckets that have a high number of objects per shard.", > >> "default": true, > >> "daemon_default": "", > >> "tags": [], > >> "services": [ > >> "rgw" > >> ], > >> "see_also": [ > >> "rgw_max_objs_per_shard" > >> ], > >> "min": "", > >> "max": "" > >> } > >> } > >> ``` > >> > >> ``` > >> > >> { > >> "option": { > >> "name": "rgw_max_objs_per_shard", > >> "type": "int64_t", > >> "level": "basic", > >> "desc": "Max objects per shard for dynamic resharding", > >> "long_desc": "This is the max number of objects per bucket index > >> shard that RGW will allow with dynamic resharding. RGW will trigger an > >> automatic reshard operation on the bucket if it exceeds this number.", > >> "default": 100000, > >> "daemon_default": "", > >> "tags": [], > >> "services": [ > >> "rgw" > >> ], > >> "see_also": [ > >> "rgw_dynamic_resharding" > >> ], > >> "min": "", > >> "max": "" > >> } > >> } > >> ``` > >> > >> > >> So when your bucket reached new 100k objects rgw will shard this bucket > >> automatically. > >> > >> Some old buckets may be not sharded, like your ancients from Giant. You > >> can check fill status like this: `radosgw-admin bucket limit check | jq > >> '.[]'`. If some buckets is not reshared you can shart it by hand via > >> `radosgw-admin reshard add ...`. Also, there may be some stale reshard > >> instances (fixed ~ in 12.2.11), you can check it via `radosgw-admin > >> reshard stale-instances list` and then remove via `reshard stale-instances > >> rm`. > >> > >> > >> > >> k > > _______________________________________________ > > ceph-users mailing list > > ceph-users@lists.ceph.com > > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com