commitlog增量开关

当我们在一个节点cassandra集群上开始性能突发运行时,日志中会出现一个“no segments i reserve;creating a new one”突发。
这可能意味着提交日志正在扩展以获取更多的段;这将影响性能指标。
以上问题:

1.Can this be avoided by pre-reserving segments for commit log and how.  
2.How can we query the current limit of a commitlog for a cassandra instace.  
3.We also notice memtable flush pointing to commitlog activity. Does commitlog sync duration  trigger table flushes  

我们使用的是3.11.x版