-
并发度参数 parallel_fragment_exec_instance_num调整过大,引发作业积压, 导致整个集群导入缓慢或直接卡死
-
排查: grep -rn “rejected” fe.log
fe.log.20230313-2:3565223:2023-03-13 21:30:08,247 WARN (thrift-server-accept|92) [ThreadPoolManager$LogDiscardPolicy.rejectedExecution():178] Task com.starrocks.common.SRTThreadPoolServer$WorkerProcess@5da069c6 rejected from thrift-server-pool java.util.concurrent.ThreadPoolExecutor@2dcce7ee[Running, pool size = 4096, active threads = 4096, queued tasks = 0, completed tasks = 204299]
pool size = 4096, active threads = 4096, 代表占用满载
- 解决办法:
调整全局parallel_fragment_exec_instance_num建议不能大于3