Shinyproxy old deployment not getting terminated

[anoop7181]

Hello,

I am deploying shinyproxy and the latest version is not getting picked automatically even with successful deployment. There are no active web socket connections or child pods present.
This was not the case before and only different was enabling promethus metrics through shinyproxy.yaml. Though it might not be the issue i am unabled find the Root Cause of this.

[LEDfan]

Hi @anoop7181

There was a bug in the operator (or more precisely, in the library we use to talk with k8s) where the operator could be confused about the state of k8s. One of the things observed was, that the operator thought there were more pods running in the cluster than in reality. This causes the operator to not remove the old shinyproxy servers. When this would have been the case, you would have seen a log statement similar too:

ShinyProxyInstance has X running apps => not removing this instance

A simple workaround is restarting the operator, but the bug should be fixed in the latest snapshot of the operator (0.1.0-SNAPSHOT-20211029.071023).