Problem: rocksdb compaction could cause write rate to drop to mimimum

[adu-crypto]

Is your feature request related to a problem? Please describe.

rocksdb compaction could cause the write QPS to drop to minimum, which slows down the node performance.

Describe the solution you'd like

rocksdb 7.5.3 said that it fixed a bug related to concurrent compactions might cause unnecessary further write stalling, and this could possibly solve this problem.
maybe we could test and upgrade rocksdb to latest version to mitigate this.
https://github.com/facebook/rocksdb/releases/tag/v7.5.3

Bug Fixes
* Fix a bug in which backup/checkpoint can include a WAL deleted by RocksDB.
* Fix a bug where concurrent compactions might cause unnecessary further write stalling. In some cases, this might cause write rate to drop to minimum.

[JayT106]

Does the concurrent compactions issue happen in v6.29.5?

I remembered the rocksDB V7 is API breaking.
It will requires modifying tm-db

[adu-crypto]

Does the concurrent compactions issue happen in v6.29.5?

it seems discussed in this issue and the solutions seem not backported.

[yihuang]

it seems not hard to make gorocksdb build with v7: cosmos/gorocksdb#7

[yihuang]

closed by #813