secrets-store-csi-driver-provider-gcp VS Reloader

The driver can also sync changes to secrets. The driver currently supports Vault, AWS, Azure, and GCP providers. Secrets Store CSI Driver can also sync provider secrets as Kubernetes secrets; if required, this behavior needs to be explicitly enabled during installation.

GCP SS-CSI driver

That's interesting actually, Google provides their own rpvider for the Secrets Store CSI Driver: https://github.com/GoogleCloudPlatform/secrets-store-csi-driver-provider-gcp

Consider: if you have a tool like terraform managing your infra components including your data layer, you likely want to manage those reaources in a different lifecycle from your application code. Applications may also likely managed using a different toolset (kubectl, helm, scaffold, etc.). In this case, secret Manager acts as the secure configuration bridge between the tools, keeping the secrets out of human hands. As certs and passwords are generated on the infra side, those values can be stored as secrets in SM. Application workloads - backed by service accounts having access to read the secret - can decrypt during launch and use the secret as needed. You can use common patterns in both GKE (via thesecrets store csi driver ) and Cloud Run for consuming secrets in this way.

I prefer https://github.com/GoogleCloudPlatform/secrets-store-csi-driver-provider-gcp

You can combine this approach with something like https://github.com/stakater/Reloader to automatically restart pods when a certain secret value changes. So if your static code needs to be rebuilt when certain values change, you can use an init container to run the build on startup.

This is not a cron job? I'm not sure if it helps, but you can have pods do a rolling restart on secret updates: https://github.com/stakater/Reloader that would clear the resources each run, but I'm not entirely clear on what you're looking to achieve.

If you use secrets as Environment Variables you will need to use something to make workloads get the new credentials, if they just loose connection. You can use the Reloader project for that.

Reloader is designed to simplify the process of updating application configurations in Kubernetes. It monitors ConfigMaps and Secrets for changes and triggers rolling upgrades for associated resources such as Deployments, StatefulSets, DaemonSets, and more. Reloader eliminates the need for manual intervention and reduces the risk of errors during the configuration update process.

The go k8s packages are pretty bloated - this may also just be a niche case. If you are looking to get secrets with hot reloading, you might also consider mounting a file or setting env vars and coupling it with this reloading operator: https://github.com/stakater/Reloader

For the auto restart this has been a staple install in all clusters for years for me: https://github.com/stakater/Reloader

If you didn't want to use SOPS, for some reason, you can certainly take advantage of external secrets as another commenter proposed, but you won't be able to accomplish (2) without an external tool adding to the mix, like Reloader: https://github.com/stakater/Reloader – that's because something has to update the HelmRelease in order to trigger it to upgrade. (You could just wait for the reconciler to come along, but the tendency is to set the polling interval longer than the default, so Helm won't be re-trying as often, in case something goes wrong...)

You can install https://github.com/stakater/Reloader And then it's just matter of a single annotation and it restarts automatically when there are changes.