ROX-15980 observability resource requests and limits#993
Conversation
|
/retest |
| resources: | ||
| requests: | ||
| cpu: 1500m | ||
| memory: 18Gi |
There was a problem hiding this comment.
Do we need 18Gi on all clusters? Perhaps the request should be smaller than the limit to accommodate smaller systems?
There was a problem hiding this comment.
I figured 18Gi would be a minimum, since for stage it currently hovers around ~10Gi with not that many centrals on it..
There was a problem hiding this comment.
But it's ok if the actual usage is above the request, right? As long as it's smaller than the limit.
There was a problem hiding this comment.
Yes, though if request < limits, it's not a guaranteed QOS and the pod might be descheduled. My initial thought on prometheus, alertmanager & so on, is that since it drives all of our alerts, I would consider this as a critical component that should have a guaranteed QOS. But it's just my personal opinion. Please feel free to suggest other values to put as requests / limits.
There was a problem hiding this comment.
Fair enough. It's a pity it takes so many resources, but I guess it is what it is.
porridge
left a comment
porridge
left a comment
There was a problem hiding this comment.
Would this fail on deployment if there's a typo somewhere here, or do we need to check by hand if the deployed values match our intention?
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: ludydoo, porridge, stehessel The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
3 participants
Sets the resource requests and limits for observability components
The values were derived from the observed metrics on prometheus/grafana.