We are using quick install script rn for deployment. For custom channel and action we pull images from our private ECR repo.
The ec2 instance is shut down daily in the night (when it is not being used).
Now, when the instance is spun up again, all pods spin up again automatically, except for those depending on our custom channel and action server images.
I understand that it may be happening as a result of the expiration of the Image Pull secret we had created to pull images from the private repo.
This wouldn’t have been such a big issue if we were able to simply create a new secret and pull those images back again using helm upgrade command.
When we run the upgrade command back again after creating a new secret, the ImagePullBackOff persists. It gets resolved only when we uninstall, do a fresh installation of the whole cluster and then pull back the images using newly created secret.
@baurjanj
could it be your rasa/rasa-sdk version not compatible? Maybe during your upgrade you should delete your namespace then check for you rasa/rasa-sdk version in this compatibility matrix table.
@stevelam15 thank you for your response, I was able to fix the issue. It was a very silly mistake on my part. I was in a wrong directory when I built the docker container for the custom actions server which picked up the wrong Dockerfile and built the wrong container. It is all good now. Thanks again!
