-
Feature
-
Resolution: Won't Do
-
Normal
-
None
-
None
-
False
-
False
-
Undefined
What is the problem that your customer is facing?
Right now it is possible to define a nodeSelector for fluentd daemonSets via .spec.collection.logs.fluentd.nodeSelector in ClusterLogging, however, more complex affinity rules such as the following is not currently possible:
{{ affinity:
nodeAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
nodeSelectorTerms:
- matchExpressions:
- key: node-role.kubernetes.io/worker
operator: Exists - key: node-role.kubernetes.io/infra
operator: DoesNotExist}}
P.S.1: This can be also added to other components (es, kibana, .. ) as well.
P.S.2: I'm willing to contribute a PR for this if the feature matches the project guidelines.What is the business impact, if any, if this request will not be made
available?
<<Please detail any impact on the 'what happens if not delivered' so that
we can better make judgement of the priority for this request.>>
What are your expectations for this feature
<<Please add any requirements you may have so that we are able to deliver
what you need.>>
Have you done this before and/or outside of support and if yes, how?
(Optional)
<<Please add any information on how you have done that so far or any ideas
you'd like to share with us.>>
- links to
[OBSDA-63] Enabling affinity rules for fluentds
Only change for vector is that the affinity section should move out of the fluentd config sub-section into the general collector section so it applies to all collector types in the same way.
That also means we can leave the existing fluentd confing section unchanged and design the new affinity section without concern for backwards compatibility.
OpenShift Jira Bot
added a comment - Roger Floren - This Epic does not yet have a priority. Please work with product management to ensure that this issue has been prioritized. Once there is a priority, this issue may be transitioned to In Progress.
OpenShift Jira Bot
added a comment - Roger Floren - This Epic does not yet have a priority. Please work with product management to ensure that this issue has been prioritized. Once there is a priority, this issue may be transitioned to In Progress. The feature is being assessed in the next team planning meeting. Updates to follow.
This request was first raised in PR https://github.com/openshift/cluster-logging-operator/issues/1105
Fluentd is deprecated in Cluster Logging going forward