FAQ for Ops - What's the default *log store duration*? How to extend the duration?

[JianfeiHu]

Organization Name:
Short summary about the issue/question:We want to keep the logs for a long time. Where do we keep the logs for the system and job? How long is the storage period?

OpenPAI Environment:
OpenPAI version:0.8.2
Cloud provider or hardware configuration:
OS (e.g. from /etc/os-release):
Kernel (e.g. uname -a):
Hardware (e.g. core number, memory size, storage size, GPU type etc.):
Others:

[mzmssg]

Job's logs are stored on hdfs, by default it will be stored under /sys/logs for 7 days.
https://github.com/Microsoft/pai/blob/9bfdbb96edc2d56b54af73c7358b6a9350fd0b3d/src/hadoop-node-manager/deploy/hadoop-node-manager-configuration/yarn-site.xml#L79

[JianfeiHu]

@mzmssg 这个值有上限吗？还有系统的日志保存在什么地方？

[mzmssg]

@JianfeiHu
Seems no limitation.
We have no a system log centralization, the general solution is getting them by docker logs .

[ydye]

For service's log, the max log size will be 100m * 10.

 "log-opts": {
    "max-size": "100m",
    "max-file": "10"
  },

[scarlett2018]

@YanjieGao @mzmssg - this seems to be an fundamental question for ops, may we update the FAQ with it?

@JianfeiHu - had you evaluated the answers and were they addressed your question?

[mzmssg]

@JianfeiHu
Refer to #2244