使用cassandra reaper 做repair的时候，执行一张20G的表就一直卡住了。
应该是token分区修复的吧，执行速度特别慢
repair线程数设置为3
日志是报这个：
postponed repair segment xxxxx because one of the hosts (xx.xx.xx.xx) was already involved in a repair
Cannot run segment xxxx for repair xxxx at the moment. Will try again later
这个能怎么解决呢？
跟The Last Pickle的对reaper很熟的专家问了一下，这是他给的解决方案：
we want to know if the segment is in a STARTED state.Please check the state of segment in the segments view in the Reaper UI.
If the segment is in a STARTED state there is a known bug in Reaper where a segment can be in a STARTED state, stalls and never completes. There is a patch coming to fix this in the latest version of Reaper. The state of the segment can be fixed in two possible ways.
Abort the repair
You can fix this issue by clicking the “Abort” button on the repair with the stalled segment.
This would terminate repairs on any segments that are yet to be completed. You would need to start a new repair which would commence repairing from the start again.
The other way to fix this is more manual, and might take a full article to explain.
完整的解决方案见附件里的PDF文件。reaper stalled issue and solution.pdf
If they are running into the STARTED segment issue, an upgrade will fix their problem.
In this case they will need to upgrade to at least 2.0.4
https://github.com/thelastpickle/cassandra-reaper/releases/tag/2.0.4
The latest release is 2.0.5
https://github.com/thelastpickle/cassandra-reaper/releases/tag/2.0.5
5 People are following this question.
