/assign @Random-Liu

@ajitak Do you expect NPD to start the docker monitor? If so, you also need to pass in the docker monitor config. https://github.com/kubernetes/kubernetes/blob/master/cluster/addons/node-problem-detector/npd.yaml#L51

@Random-Liu updated yaml to pass docker monitor config.

@ajitak We should think about this more...

Maybe we should only enable this on gci, because log parsing introduces extra overhead, especially when the log is quite spammy. Docker log may be quite spammy if the log level is too low, I'm not sure whether we want to enable docker log parsing everywhere by default.

Maybe only enable on gci and only for standalone mode which is used by GKE? https://github.com/kubernetes/kubernetes/blob/master/cluster/gce/gci/configure-helper.sh#L977

/cc @dchen1107 WDYT?

Fixes #47219.

@kubernetes/kubernetes-release-managers This helps fix a 1.7 issue.

The memory overhead looks ok by design based on the data published at: kubernetes/node-problem-detector#85 (comment)
since the new change in new NPD image only introduce a new goroutine for docker logs, and by design we bound the memory per goroutine.

@ajitak Can you check the log level of that docker image corruption issue at docker log? If the docker level is info, not warning or above, the new config in NPD doesn't help the production.

@dchen1107 log level of docker image corruption issue at docker log is error.

Great!

/lgtm

Offline discussed with @dchen1107 and @ajitak. The extra resource overhead should be fine based on kubernetes/node-problem-detector#85 (comment).

We should pay attention to the node resource usage after this PR to make sure the docker log parsing doesn't introduce too much overhead.

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: ajitak, dchen1107

Associated issue: 47219

The full list of commands accepted by this bot can be found here.

Automatic merge from submit-queue (batch tested with PRs 47993, 47892, 47591, 47469, 47845)

@ajitak You need to cherry-pick this to 1.7. See https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md.

Commit found in the "release-1.7" branch appears to be this PR. Removing the "cherrypick-candidate" label. If this is an error find help to get your PR picked.