Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add tolerations for Stackdriver Logging and Metadata Agents. #69737

Merged
merged 1 commit into from
Oct 15, 2018
Merged

Add tolerations for Stackdriver Logging and Metadata Agents. #69737

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
5 changes: 5 additions & 0 deletions cluster/addons/fluentd-gcp/fluentd-gcp-ds.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,11 @@ spec:
nodeSelector:
beta.kubernetes.io/fluentd-ds-ready: "true"
terminationGracePeriodSeconds: 60
tolerations:
- operator: "Exists"
effect: "NoExecute"
- operator: "Exists"
effect: "NoSchedule"
volumes:
- name: varlog
hostPath:
Expand Down
5 changes: 5 additions & 0 deletions cluster/addons/metadata-agent/stackdriver/metadata-agent.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,11 @@ spec:
restartPolicy: Always
schedulerName: default-scheduler
terminationGracePeriodSeconds: 30
tolerations:
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Discussion topic: do we need similar tolerations in the cluster-level agent spec? I suspect the only reason we would is if all nodes are tainted. @x13n @MaciekPytel @kawych @piosz @bmoyles0117

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Some of that conversation happened in #68920 (comment).

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah, thanks for the pointer. I see the priority class now. Given that we'll only cherry-pick this into 1.12 and 1.11, this should work. If we ever need a 1.10 cherry-pick, we'd have to add stronger tolerations (e.g., CriticalAddonsOnly).

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

CriticalAddonsOnly isn't really stronger - the ones added here are wildcard ones, so they include it already.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I meant for the cluster-level agent, which currently doesn't have any tolerations.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh, right. I think if someone is tainting all their nodes, something may break anyway, but yes, this could at least mitigate the problem in 1.10.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think a lot of things won't schedule if you taint all your nodes (DNS, metrics-server, ...). I don't think cluster-level agent should have stronger tolerations than other cluster-level system components.

- operator: "Exists"
effect: "NoExecute"
- operator: "Exists"
effect: "NoSchedule"
updateStrategy:
rollingUpdate:
maxUnavailable: 1
Expand Down