Giters

openshift / cluster-logging-operator

Operator to support logging subsystem of OpenShift

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Feature request: Enable podTemplate for elasticsearch

YiannisGkoufas opened this issue · comments

commented

Hi all, I am hitting this issue: https://access.redhat.com/solutions/5195611
Basically I need to be able to add this configuration:

podTemplate:
  metadata:
    labels:
      tuned.openshift.io/elasticsearch: ""

on the elastricsearch part of the spec of ClusterLogging but I don't think it's supported.
Let me know if the functionality is there and I am missing something.

commented

Please describe the use-case where this needs to be propagated to the pod deployments? Why would we not add it to all deployments?

commented

As described in here: https://access.redhat.com/solutions/5195611 when I try to deploy ClusterLogging, elasticsearch fails to start with the error:
max virtual memory areas vm.max_map_count [65530] is too low, increase to at least [262144]
The resolution described is to update the podTemplate spec of elasticsearch.

How can I add it to all the deployments?

commented

Thanks for the immediate response @jcantrill

commented

This label is added as part of the node tuning operator that adjusts worker node settings to account for the virtual memory changes you identify. If you are deploying logging on a RedHat supported cluster then all the mechanisms should already be in place to address this issue. If you are attempting to deploy on the "origin" or upstream I can not provide any advice at this time. All the CI, etc is geared towards deploying and testing official OpenShift deployments. There is nothing in place to test otherwise

commented

I am using Openshift 4.6 provisioned in IBM Cloud and followed the instructions provided here https://docs.openshift.com/container-platform/4.6/logging/cluster-logging-deploying.html and these settings are not applied.
In addition I can't edit the deployment,pods etc since they are controlled by the operator.
The workaround was to add the label in this part https://github.com/openshift/elasticsearch-operator/blob/master/internal/elasticsearch/common.go#L322

commented

@jeremyeder Who owns the node-tuning-operator and might understand what may be at issue here?

commented

@YiannisGkoufas closing this issue. I encourage you to engage with this team to understand what may be required to incorporate the changes required https://github.com/openshift/cluster-node-tuning-operator