Nil pointer in manager.go causing a docker daemon restart

[0x0elliot]

Hey,

One of our users hosting Shuffle locally had downtime. Upon digging deeper, we noticed that the docker daemon had restarted (which led to our docker services restarted BUT our docker compose containers not restarting). Regardless, the downtime wasn't expected.

Here are the journalctl logs pointing to the nil pointer code:

Here are the client and server versions:

It seems like the issue is happening somewhere around here:

swarmkit/manager/manager.go

Lines 1123 to 1127 in 3a23580

	go func(orchestrator *replicated.Orchestrator) {
	if err := orchestrator.Run(ctx); err != nil {
	log.G(ctx).WithError(err).Error("replicated orchestrator exited with an error")
	}
	}(m.replicatedOrchestrator)

For more context, we have docker compose "services" and a bunch of docker swarm services running in this environment.

I don't mind helping fix the issue if you can give me some directions. This affects our customers directly, and I love OSS.

[0x0elliot]

Closed by #3223