doris 规模 3FE + 3BE
运行一段时间后出现 insert 需要10S 但是select 很快 ms级别就好
show load里面 确实有很多insert 但是 从磁盘的性能来看 没有很高的写入操作,日志中有大量的18 task_worker_pool.cpp:730] publish version error, retry. [transaction_id=95248511, error_tablets_size=225]
W0829 11:30:53.148705 24515 task_worker_pool.cpp:743] publish version failed. signature:95295585, error_code=-914
W0829 11:30:53.149102 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890689 txn id 95229017
W0829 11:30:53.149116 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890693 txn id 95229017
W0829 11:30:53.149119 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890697 txn id 95229017
W0829 11:30:53.149122 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890701 txn id 95229017
W0829 11:30:53.149125 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890705 txn id 95229017
W0829 11:30:53.149127 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890709 txn id 95229017
W0829 11:30:53.149130 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890713 txn id 95229017
W0829 11:30:53.149133 24515 engine_publish_version_task.cpp:72] could not find related rowset for tablet 69890717 txn id 95229017
查看 profile: 10s 很奇怪
Execution Summary:
- Analysis Time: 13.542ms
- Plan Time: 164.319us
- Schedule Time: N/A
- Wait and Fetch Result Time: N/A
Execution Profile afe7ca73f9c94adc-86c182b0a98fa2e7:(Active: 10s98ms, % non-child: 100.00%)
监控平台doris 的各项指标 不能很明显的看到异常,有运维大佬指导吗