[jboss-jira] [JBoss JIRA] (SWSQE-294) Investigate ElasticSearch memory requirements on B24 cluster

[Kevin Earls (JIRA)]

    [ https://issues.jboss.org/browse/SWSQE-294?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13591007#comment-13591007 ] 

Kevin Earls commented on SWSQE-294:
-----------------------------------

Hi [~gbaufake] . I'll attach a log, but I'm not sure it will help.  The one I'm attaching is from a node that failed after 1 of the other nodes got the OOMKilled action.  I was doing a "oc logs -f" on all nodes, but the problem is that when a node is killed it just terminates.  There's nothing of interest in the log.  I'm guessing we'd need to figure out who is killing it and look at their log.

You might want to try creating a temporary project and doing the curl and oc create steps above, that may give you more info.


> Investigate ElasticSearch memory requirements on B24 cluster
> ------------------------------------------------------------
>
>                 Key: SWSQE-294
>                 URL: https://issues.jboss.org/browse/SWSQE-294
>             Project: Kiali QE
>          Issue Type: Bug
>            Reporter: Kevin Earls
>            Assignee: Guilherme Baufaker Rêgo
>            Priority: Minor
>         Attachments: es1.log
>
>
> On minishift and the CNCF CI OpenShift cluster, we install ElasticSearch using the commands:
>     curl https://raw.githubusercontent.com/RHsyseng/docker-rhel-elasticsearch/5.x/es-cluster-deployment.yml --output es-cluster-deployment.yml
>     oc create -f es-cluster-deployment.yml
> However, on B24 this fails.  The first 2 elasticsearch pods will start up, but then elasticsearch-2 gets into the following failure loop like:
> elasticsearch-2   1/1       Running   3         1m
> elasticsearch-2   0/1       OOMKilled   3         1m
> elasticsearch-2   0/1       CrashLoopBackOff   3         2m
> If I increase memory in the es-cluster-deployment.yml to 1Gi it will at least start, but I've found I really need to increase it to 2Gi to do anything non trivial.  
> We may need to track down the people who originally created this template for help on this. 



--
This message was sent by Atlassian JIRA
(v7.5.0#75005)