Help requested on self-inflicted DNS issue

[nprampage]

Hey, fellow Pihole users. I'm looking for some insight into an issue I encountered yesterday where Pihole stopped resolving DNS queries after a router reboot until I manually restarted my Raspberry Pi.

For the devs, if they're feeling kind enough to peek, by debug token is https://tricorder.pi-hole.net/0C6KOCj4/.

My network uses a Ubiquiti UDM-SE device as the gateway and router. I saw yesterday around 2pm central time that it had been up and running for several weeks without a reboot, and since no one was using the network I decided to run a restart for preventative (hah!) maintenance.

When it came back up, all appeared okay from my remote location, however the wireless access points were reporting connectivity issues. The cause of this wasn't immediately clear; suffice it to say I performed a lot of pointless chasing of the wrong thing until I could get onsite to realize what had happened.

My gateway and Internet hardware have DNS entries for public DNS servers. My LAN has DHCP assignments to use the Pihole. So my Internet hardware was fine, but everything on my LAN (including wireless devices) was pointing to my Pihole. The Pihole was in a bad state; I could pull up the Web interface, but all the dashboard stats were empty or showing indications of an error.

Once I saw the Web interface, I went to System in Pihole and restarted DNS. This didn't help, so I clicked on the Restart System button. A couple of minutes later, everything was working fine.

So clearly this is due to something I initiated. What I'd like to do is understand why Pihole didn't recover once the Internet was again available to my devices after the router finished rebooting.

For reference, the reboot was issued on May 18 14:39 in the logs, which is when you begin to see a bunch of SERVFAIL responses. Obviously that's expected, but the restart took less than five minutes. Recovery didn't actually occur until about 2 hours later when I was able to restart the R-Pi.

Is there something I could have done to avoid this from a setup standpoint? Or was it just a weird glitch/coincidence?

Thanks in advance for any feedback you all might have.

[nprampage]

Upon further investigation in the query logs, I see the vast majority of queries during the time window mentioned gave an N/A response. Since Unbound is my DNS Resolver, I guess that means Unbound was upset. Perhaps I could have corrected the issue simply by restarting the Unbound service instead of a full R-Pi restart?

[rdwebdesign]

@nprampage

Is there something I could have done to avoid this from a setup standpoint? Or was it just a weird glitch/coincidence?

I guess your debug log was generated AFTER the recovery. It shows a normal Pi-hole, without issues.

I can't say for sure what happened, but my personal guess is: this was a weird glitch. If this happens again, try to generate a debug log while the problem occurs (run pihole -d from the command line).

[Bucking_Horn]

The most obvious cause for unbound failing would be wrong time information.
However, I'd expect consisent SERVFAILs then.

Nevertheless: Could your router reboot have caused incorrect time information on your RPi hosting Pi-hole/unbound?

Since you've observed N/As instead:
Would your logs indicate that unbound has been inaccessible/unavailable during your observed outage?

[nprampage]

I have no logging enabled on Unbound, so there aren't any leads there that I can find. Lesson learned there, I suppose. My router had the correct time before and after the reboot, since it's upstream DNS is set to Quad9.

I've restarted the router several times in the past, so there was something unique that happened here causing the R-Pi to become upset. I'm just trying to figure out if it's something I can prevent in the future. It may just have to remain a mystery.