Flakey DHCP implementation?

I have a gateway machine that is also the DHCP server for the network.

I notice that if I power off this machine (or even power cycle it for a kernel update), rockstor becomes unreachable very quickly - everything else on the network remains pingable for a long time, but rockstor loses connectivity far faster, and also does not recover (the key part). If I unplug the ethernet connection and replug it again connectivity comes back.

I’d like my nas to be more resilient to this.

Before I troubleshoot this some more - does anyone else have a similar problem? Is this a known bug?


I have the same experience with DHCP but the service was up all the time.
After a while the rockstor box was unreachable. I have no screen to it but a simple press on the powerbutton should take it down smoothly. In the case of the “DHCP” issue the box did not shutdown when i press the powerbutton so i know it was stalled in some way.

This was driving me crazy and i even bought new motherboard, cpu, memory in case my old Atom D525 was the cause but same results.
I switched over to fixed IP 7 days ago and it’s been stable after that. I did dig into logs trying to understand when it was breaking and from what i can see the last thing it logs is docker network changes.

When i had DHCP enabled it was failing and needed hard reboot 1-2 times a day and now with fixed IP it’s stable for 7 days.