Results 9 comments of Beorn Facchini

Fair enough. I moved to golang.org/x/crypto/acme around this time.

This PR isn't stale. I'll look into the implementation at a future date.

A possible solution to the bit shifting bug above without repeated calculations or magic values is below: ```go waitFor = baseRetryInterval maxRetryInterval || waitFor < baseRetryInterval { waitFor = maxRetryInterval...

I have found a third bug that is due to missing internal state in volume publishing and unpublishing, when controller client is unset here [permalink](https://github.com/moby/swarmkit/blob/ea1a7cec35cb05dce3cff2de3f4ee28cbca6ca79/manager/csi/plugin.go#L203-L246): ```go // PublishVolume calls ControllerPublishVolume...

I have found a fourth bug that is due to a race condition and is a little harder to reason about. It happens when the `csi-test` service is running with...

You can find examples in Hetzner documentation on creating a volume and using it in a service here: https://github.com/hetznercloud/csi-driver/tree/main/docs/docker-swarm. There's also a repository with examples of using various Docker Swarm...

I can confirm that I saw this swarm bug with a CSI volume plugin and forgot to report it. I am currently using Docker v26.0.2 which doesn't have this regression.

Similarly, this early return in `NodeUnpublishVolume` skips the log which would be helpful in debugging: https://github.com/moby/swarmkit/blob/ea1a7cec35cb05dce3cff2de3f4ee28cbca6ca79/agent/csi/plugin/plugin.go#L366-L373