[jboss-jira] [JBoss JIRA] (SWSQE-200) B12 OpenShift Cluster is flaky

[Kevin Earls (JIRA)]

    [ https://issues.jboss.org/browse/SWSQE-200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13575497#comment-13575497 ] 

Kevin Earls commented on SWSQE-200:
-----------------------------------

Hi [~gbaufake] . This is much better.  I can now run the Flaky job repeatedly, and have been able to run the standalone performance test with both Cassandra and ElasticSearch.  

There are a couple of minor issues remaining:

-- I was not able to get a persistent Jenkins instance running in the "jeager-test" project, so for expedience sake I created jaeger-test2 and created an ephemeral jenkins.  I tried to delete jaeger-test, but it's still around, so you might be able to find some clues there on why that's failing.

-- I still have to increase the memory for ElasticSearch to 1G even though it runs elsewhere fine with 512M.  

I can open new JIRAs for these if you'd like, or we can leave this open but lower the priority.



> B12 OpenShift Cluster is flaky
> ------------------------------
>
>                 Key: SWSQE-200
>                 URL: https://issues.jboss.org/browse/SWSQE-200
>             Project: Kiali QE
>          Issue Type: QE Task
>            Reporter: Kevin Earls
>            Assignee: Guilherme Baufaker Rêgo
>            Priority: Blocker
>
> I'm opening this mostly as a placeholder, and will update it as I get more information on the problems I've been experiencing.  Since I've been using B12 I've been experiencing more frequent failures than on other clusters, including minishift on my laptop and the CNCF CI Jenkins.  Here are a couple of instances:
> 1.  Deploying ElasticSearch along with the Jaeger Production templates requires allocating 2G or memory, even though the default 512M works fine elsewhere.  This can be seen in the Jaeger Standalone Performance tests job here: https://jenkins-jaeger-test.openshift3.jonqe.lab.eng.bos.redhat.com/job/Jaeger%20Standalone%20Performance%20Test/ . If you run the job without changing the ES_MEMORY parameter to 2Gi it will fail.
> 2. I have a set of smoke tests for Red Hat productized artifacts for the Jaeger Java client which are run using the Jaeger all-in-one template.  It's fairly simple, but on B12 deployment of the Jaeger all-in-one images fails on every other build.  I've made a copy of the job here:  https://jenkins-jaeger-test.openshift3.jonqe.lab.eng.bos.redhat.com/job/Flaky%20Test/ and its history should be clear.
> So far I have not been able to get any useful information about why this is failing.  In the OpenShift console it looks like Jaeger has started correctly.  There are no errors in the logs, nor any to be found under monitoring.  But if you click on the Jaeger link, you get the message "Application is not available . The application is currently not serving requests at this endpoint. It may not have been started or is still starting."



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