You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@linkis.apache.org by zhaoyuan <15...@163.com> on 2022/04/29 07:27:06 UTC

[DISCUSSION] Query invalid service

hcl:

微服务情况下,在linkis-appjointentrance重启后,linkis_task 表记录了实际的ip:port。导致一直在向失效的服务做查询。导致平台卡顿。

@peacewong@WDS 这块有优化吗




叛逆的守护者:

目前看是卡在Failed to obtain lock:linkis-cg-engineconnmanager这里了加减ecm资源




peacewong@WDS:

这块有优化的,做了一个一键重置任务的脚本,服务stop或者异常的时候执行下这个脚本。




peacewong@WDS:

那这里确实是获取锁卡住了,但是可能获取锁卡住的真正原因应该是资源加减那里,得看看这个锁被谁持有了,然后做了什么耗时操作。


hcl:
In the case of microservices, after the restart of linkis appjointratance, linkis_ The task table records the actual IP: port. As a result, queries have been made to failed services. Cause the platform to jam.
@ peacewong@WDS   is this piece optimized


Rebellious Guardian:
At present, it seems that it is stuck in failed to obtain lock: links CG engineconnmanager, adding and subtracting ECM resources


peacewong@WDS :
This is optimized. A one click Reset task script is made. This script is executed when the service stops or exceptions occur.


peacewong@WDS :
It is true that the acquisition lock is stuck here, but the real reason why the acquisition lock may be stuck should be the addition and subtraction of resources. It depends on who holds the lock and what time-consuming operations are done.