[ https://issues.apache.org/jira/browse/CASSANDRA-10099?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15130534#comment-15130534 ]
Marcus Eriksson commented on CASSANDRA-10099: --------------------------------------------- [~yukim] agreed, but I think we should do both After CASSANDRA-6696 and especially CASSANDRA-10540 we can have very many compaction strategy instances (one per owned vnode range, typically 768 with rf=3 and 256 tokens), so even if we don't loop, this can take a long time. But since each instance is totally independent from the other instances we don't need the global synchronization - we just need to make sure that noone switches out the compaction strategy under our feet. I'll push an updated branch with both approaches unless you disagree? > Improve concurrency in CompactionStrategyManager > ------------------------------------------------ > > Key: CASSANDRA-10099 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10099 > Project: Cassandra > Issue Type: Improvement > Reporter: Yuki Morishita > Assignee: Marcus Eriksson > Fix For: 2.1.x, 2.2.x, 3.x > > > Continue discussion from CASSANDRA-9882. > CompactionStrategyManager(WrappingCompactionStrategy for <3.0) tracks SSTable > changes mainly for separating repaired / unrepaired SSTables (+ LCS manages > level). > This is blocking operation, and can lead to block of flush etc. when > determining next background task takes longer. > Explore the way to mitigate this concurrency issue. -- This message was sent by Atlassian JIRA (v6.3.4#6332)