Conjur pod restart fails following auto account create

[diverdane]

Summary

When a Conjur cluster is Helm installed using the auto account create feature e.g. with this chart setting:

    --set account.create=true

and then the Conjur server pod crashes before any helm install ... operation has been done,
then the Conjur pod will continually crash and restart:

$ kubectl get pods
NAME                          READY   STATUS             RESTARTS   AGE
conjur-oss-555f6b6f6b-bms88   1/2     CrashLoopBackOff   1          19s
conjur-oss-postgres-0         1/1     Running            0          103s
$

When this occurs, the Conjur server logs show
an error similar to the following:

Account 'default' already exists
error: exit

Steps to Reproduce

Steps to reproduce the behavior:

1. Helm install a Conjur-OSS cluster using the --set account.create=true command line argument.
2. Get the Conjur pod name with kubectl get pods -n <conjur-namespace>
3. Reset the Conjur pod with kubectl delete pod -n <conjur-namespace> <conjur-pod-name>
4. Use kubectl get pods -n <conjur-namespace> and see that Conjur pod is continually crashing and restarting.
5. View Conjur server logs with kubectl logs -n <conjur-namespace> <conjur-pod-name> conjur-oss.

Expected Results

A clear and concise description of what you expected to happen.

Actual Results (including error logs, if applicable)

Restarting the Conjur pod manually should bring up the server successfully, with no crashes.

Reproducible

• Always
• Sometimes
• Non-Reproducible

Version/Tag number

Version 2.0.2

Environment setup

Any Kubernetes platform.

Additional Information