3.1 存算分离stream load compaction

logstash stream load 导入到 3.1 存算分离集群报错：
txn_id=3018936512093184, node=10.10.25.104:8060, add batch req success but status isn't ok, err: [E-235]PStatus: (10.10.25.104)[E-235]failed to init rowset builder. version count: 20001, exceed limit: 20000, tablet: 1760743536409. Please reduce the frequency of loading data or adjust the max_tablet_version_num or time_series_max_tablet_version_numin be.conf to a larger value
根据描述看应该是导入频率高，但是compaction不及时导致的，尝试调整了 max_tablet_version_num 但只是缓解，等会还是会出现，说明 compaction 速度已经跟不上了， 请问下调整哪些配置能够提升compaction速度？