litmus-helm icon indicating copy to clipboard operation
litmus-helm copied to clipboard

Cannot re-install litmus-agent after uninstalling it

Open Calvinaud opened this issue 2 years ago • 1 comments

Hello,

I have a problem where we cannot re-install the litmus-agent after un-installing it. When trying to re-install the agent the subscriber and event-tracker pod are failing with the following error: level=fatal msg="required key VERSION missing value" or level=fatal msg="required key SERVER_ADDR missing value". The reason is that the configmap (litmus-config) and the secret (litmus-agent) are not populated by the pre-hook job during the reinstall.

I think the origin of this problem is that during the uninstallation of the litmus-agent, the chaos-agent is not removed from the chaos center. Which mean the pre-hook job think that the litmus-agent is already installed on the chaoscenter and so skip the part where it add information to the configmap and the secret.

As a quickfix, removing the litmus-agent manually from the chaoscenter solve the issue and it's possible to re-install the litmus-agent. As a long term fix, my first idea would be to a "pre-delete" job that remove the litmus-agent from the chaoscenter. Any opinion on this idea is welcome.

Have a nice day.

Calvinaud avatar Sep 15 '23 14:09 Calvinaud

Hi @Calvinaud , Thanks for bringing this up. We need to document the steps to uninstall the agent first from chaos center and then do the uninstallation of helm-agent in the cluster.

uditgaurav avatar May 13 '24 11:05 uditgaurav