Mark called out [0] a specific issue [1] in older versions of rocksdb that
may cause problems when running multiple concurrent compaction threads.
The luminous 12.2.13 rocksdb submodule branch rests somewhere between the
following rocksdb tags:
tags/v4.9 (960 commits ahead)
tags/v5.5.1 (97 commits behind)
The concurrency issue was fixed by a commit [2] that landed well after
v5.5.1:
git describe --contains 4420df4b0e15ae88911e960c4fbafbaf8450fcf7
v5.15.10~154
The fix is present in nautilus, but not in luminous.
Am I missing something that mitigates the risk of running concurrent
compactions in Luminous? Should this default be reverted back to 1?
[0]
https://github.com/ceph/ceph/pull/29027#issue-297158998
[1]
https://github.com/facebook/rocksdb/pull/3926
[2]
https://github.com/facebook/rocksdb/commit/4420df4b0e15ae88911e960c4fbafbaf…
Thanks,
Dan Hill