Previously, we would run N concurrent compactions if there were N levels requiring compaction. Now we only run concurrent compactions for high priority compactions where the priority is given by the compaction score. The high priority threshold is set to 1.5 which will cause intra-L0 compactions to be considered high priority and any non-L0 compaction where there is a significant level size imbalance.
Fixes #47
该提问来源于开源项目：cockroachdb/pebble
