ROX-15980 set resource requests and limits to the egress-proxy #991
Conversation
| limits: | ||
| cpu: 100m | ||
| memory: 128Mi | ||
| requests: |
There was a problem hiding this comment.
I think you got the request and limit values backwards.
|
/retest |
| egressProxy: | ||
| image: ubuntu/squid:5.2-22.04_beta | ||
| replicas: 2 | ||
|
|
||
| replicas: 1 |
There was a problem hiding this comment.
this doesn't need to be fixed in this PR, but this does make me wonder if we should actually run two replicas in the cloud service for all the usual reasons.
| cpu: 100m | ||
| memory: 128Mi | ||
| limits: | ||
| cpu: 200m | ||
| memory: 256Mi |
There was a problem hiding this comment.
How about we use requests==limits at least until we have more replicas? We'll likely be setting resources for a bunch of pods in the next rollouts (see sibling PRs) so the risk of facing evictions will be higher than normal and I'm a bit worried with one replica this may cause service degradation for non-trivial central configurations.
There was a problem hiding this comment.
I can add the replicas in this PR perhaps?
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: kurlov, kylape, ludydoo, porridge The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Sets the resource requests and limits for the egress-proxy.
The values were derived from the observed metrics on prometheus/grafana. These resource defaults are a bit overkill for the actual observed usage. I was not comfortable putting less than this for a production deployment.
It also seems like the
values.yamlfile was ignored for thetenant-resources. This PR also adds changes to use these values by default and apply overrides on top of them.The current cluster configuration deploys 1 replica of the
egress-proxy, so I've changed the value to reflect that as well.