I've been feeling the pain from Bitnami's decision to sunset their images. It seems like every time I try to re-deploy Postgres and Redis on my environments—production, staging, and development—I run into these endless ImagePullBackOff loops. After a lot of trial and error, I've decided to switch to CloudNativePG for Postgres and stick with Bitnami's legacy version for Redis just to keep things running smoothly. Has anyone else found a better migration path or a solid replacement solution?
5 Answers
You might consider hosting your own mirror if you can. It could save you some headaches with the transitions. We use blue/green deployments for environments which helps mitigate issues. Just be wary of costs with spot instances for staging, though.
We switched to the legacy image repository a couple of months back, and now we're slowly replacing those images with alternatives. So far, we’ve avoided downtime. It’s a hassle, but at least we had some warning to make changes.
You should definitely have CICD pipelines for canary builds and infrastructure testing. Plus, setting up Dependabot can help you catch these issues before they become a problem in production.
We recently moved to the Redis Operator from OT Container Kit for Redis. It’s been a smoother experience for us! You might want to check it out if you haven't already.
I’m a fan of CloudNativePG too! I haven’t regretted switching. We’re also working on migrating other services like RabbitMQ and Mongo. It's a process but worth it in the long run.
We utilize blue/green setups as well. It's been helpful, especially when we have to pull the latest version of Postgres. Just make sure your instances are scaled correctly to minimize costs.