Admin user repeatedly asked to login / redirect loop

[testerUnito]

My admin user is repeatedly being asked to log in. I am redirected to an error message like the one below however it never succeeds. I believe it relates to an XSRF token expiry. How do I debug this issue.

Request URL: /secure/admin/WebSudoAuthenticate.jspa

We have captured the submitted values (shown below). Please copy it now. This text will be lost when you leave this screen.

• webSudoDestination

  /secure/admin/ViewApplicationProperties.jspa

• webSudoIsPost

  false

Please log in and retry this action.

I am using v1.21 of the atlassian-data-center helm chart (https://atlassian.github.io/data-center-helm-charts/).

[Yevhen]

@testerUnito you need to configure session affinity aka sticky cookie for your ingress. 

[Jim Knepley - ReleaseTEAM]

I suspect you need to update some settings in server.xml because the external request looks different enough from what eventually reaches the service.

https://confluence.atlassian.com/kb/cross-site-request-forgery-csrf-protection-changes-in-atlassian-rest-779294918.html

I had a similar issue with Bitbucket DC earlier today where the URL wasn't the same outside of the load balancer vs inside. The server.xml changes suggested in that article should do the trick.

[testerUnito]

Thanks Jim. Is it possible to configure the server .xml from the helm chart? My understanding is that helm is the only supported way to deploy Jira Dc on-premises.

I was able to resolve this by scaling down to 1 replica and removing the Pod Disruption Budget., however that means that my instance is not highly available.

[Jim Knepley - ReleaseTEAM]

I've never had any problem getting support assistance while running Jira DC on EC2 instances in AWS.

You can customize configuration files like this:

https://confluence.atlassian.com/jirakb/customizing-jira-configuration-files-within-a-kubernetes-environment-via-values-yaml-1277625311.html