Reboot Allocation

[Smithx10]

Currently within the plugin architecture there is no way to reboot an instance in place. I'd like a way to reboot an allocation in place with an exponential back off. This would be similar to the container restartPolicy within Kubernetes Pod Lifecycle. https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle/#restart-policy.

The idea is that by restarting the allocation, or sending a signal, someone could restore the allocation health at a cheaper cost than a full killing and restarting of a task. This would advantageous when paired with a service health check.

I'm new to Nomads ecosystem and perhaps have implemented Restart Incorrectly, but I don't think so.

[endocrimes]

👋 @Smithx10,

Thanks for your issue, we're actually working to improve some of the control around this in 0.9.2. However, Nomad has a few pre-existing ways that will restart allocations:

• If configuration or secrets change consul-template can trigger a restart or signalling of your task
• if you register service health checks with Consul, we'll handle restarting tasks when the health checks fail
• If your application crashes, or otherwise terminates, we'll restart it according to your restart stanza

And in Nomad 0.9.2, we'll be introducing support for manual restarts and signalling of tasks, as implemented in:

• nomad alloc restart alloc-lifecycle: nomad alloc restart #5502, docs: Add docs for nomad-alloc-restart #5565
• nomad alloc signal allocs: Add nomad alloc signal command #5515, Docs for nomad alloc signal #5612
• nomad alloc stop alloc-lifecycle: nomad alloc stop #5512, Documentation for nomad alloc stop #5597

However, we don't currently support the delay_function from our rescheduling logic in local restarts. If that's a specific feature that you'd like to see added, I'd welcome another issue for it.

Thanks again!

[github-actions]

I'm going to lock this issue because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.