secrets-store-csi-driver VS sops

I'm not a fan of this approach. I think the Secrets Store CSI Driver (https://secrets-store-csi-driver.sigs.k8s.io/) has a better approach.

Secret Store CSI Driver is what we're playing with now. Pretty excellent.

If you deploy on k8s, keep your eye on https://secrets-store-csi-driver.sigs.k8s.io/

Considering the major limitations of using Kubernetes Secrets, there are many new approaches being developed by the Kubernetes community. Kubernetes SIGs like the Secrets Store CSI Driver and solutions like the external secrets operator that works with third-party secret managers, and options to seal secrets through tools like bitnami’s sealed-secrets. To skip the tools and move directly to best practices, click here.

Just to clarify, CSI secret driver is from cncf not Microsoft. Only msft piece is the portion that integrates with key vault. https://secrets-store-csi-driver.sigs.k8s.io/

https://secrets-store-csi-driver.sigs.k8s.io/ and https://developer.hashicorp.com/vault/tutorials/kubernetes/kubernetes-secret-store-driver

The Secrets Store CSI Driver is a native upstream Kubernetes driver that can be used to abstract where the secret is stored from the workload. If you want to use a cloud provider's secret manager without exposing the secrets as Kubernetes Secret objects, you can use the CSI Driver to mount secrets as volumes in your pods. This is a great option if you use a cloud provider to host your Kubernetes cluster. The driver supports many cloud providers and can be used with different secret managers.

If you want security they are both bad, use something like the secret manager of your choice API directly in your app or https://secrets-store-csi-driver.sigs.k8s.io/ this will keep the actual secrets out of etcd and env vars and give you more security

My script just sets up default .sops.yaml for https://github.com/getsops/sops

You can further edit .sops.yaml(eg have multiple of them) and decide how you split secrets in your directory tree to further customize who can decrypt the secrets.

It works pretty well for prod/dev splits, etc

Mozilla SOPS (Secrets OPerationS) is an open-source command-line tool for managing and storing secrets. It uses secure encryption methods to encrypt secrets at rest and decrypt them at runtime. SOPS supports a variety of key management systems, including AWS KMS, GCP KMS, Azure Key Vault, and PGP. It's particularly useful in a DevOps context where sensitive data like API keys, passwords, or certificates need to be securely managed and seamlessly integrated into application workflows.

Encrypted secrets thanks to SOPS and Age

We do the exact same thing to keep track of some credentials we use sops[1] and AWS KMS to separate credentials by sensitivity, then use the git differ to view the diffs between the encrypted secrets

Definitely not best practice security-wise, but it works well

[1] https://github.com/getsops/sops

For anyone new to SOPS like I was - https://github.com/getsops/sops

I've found this is largely solved by strictly separating plain config and secrets, and then having secrets pull from GCP secret manager / vault / whatever.

You can then commit all the config (including the secret identifiers) and it all just works so long as you're authenticated with your secret storage system.

We do this for the live configuration as well in line with Gitops and find it to work well.

If you don't want to use a cloud secret manager you can also use something like https://github.com/getsops/sops to commit the encrypted secrets safely

Basically, the simpler the better --just encrypt your secrets and check them in to version control.

We use SOPS[0] for this, and have found it to be pretty nice.

[0]: https://github.com/getsops/sops

The answer is that secrets shouldn't be stored in the git repo at all, but somewhere safe like a password manager or Mozilla's SOPS which people seem to love.

Unfortunately, the SOPS project is in some sort of a limbo state and there has been quite a long period with limited maintenance and unclear position from Mozilla. Despite the project being accepted into the CNCF, it's still unclear what will happen with it going forward.