Hide Forgot
Large bucket indexes represent a scaling issue if customer don't shard the bucket ahead of time. If the bucket index grows too large, performance can suffer. To ensure customers don't create too large a bucket index, we should set a default quota of 100k by default. This assumes the customer is not sharding by default. If they do shard, we can recommend (in the docs) that they set the bucket index quota to be 100k * the number of shards. However, the customer will explicitly have to change the default cap to do.
Why not just have a quota of 100k * #shards in code, and be done? Is it really a good idea to add another "shoot yourself in the foot" option to the conf file?
PR for this issue in https://github.com/ceph/ceph/pull/11711
The combination of https://bugzilla.redhat.com/show_bug.cgi?id=1391500 and https://bugzilla.redhat.com/show_bug.cgi?id=1391650 cover the functionality that this BZ was trying to accomplish for 2.1 for both new and existing installs.