Have been running PiHole docker on Unraid along with unbound for years without issue. I keep the docker containers updated regularly. After deciding to update from Unraid 6.8 to 6.12.4, I began experiencing frequent server crashes. Isolated the problem to PiHole before reading unraid forum to become aware of all the macvlan call trace issues with 6.12.x. PiHole and Unbound were my only containers previously running on br0 with custom IPs. I followed unraid 6.12.4 release note instructions, changed docker custom network type to ipvlan and started running PiHole and unbound with custom IPs on eth0. No more server crashes and no errors in unraid logs! Problem solved so I thought.. see below for new issue.
Actual Behaviour:
Now after making above changes, things seemed to be working as usual, but every so often throughout the day, traffic outside my local network will cease for ~15 seconds to a couple minutes. I will attempt to load a website and it will just hang. During that time, I can still access anything on my LAN. Can access unraid, no log errors or significant resource utilization. Can access PiHole WebUI. No errors in log. Ultimately, within seconds to a couple minutes, everything is back to normal and working as expected until the next time it happens again. Have never seen this before the updating unraid OS mess. Tried removing unbound temporarily and using google as my upstream DNS, still same issue. Any suggestions much appreciated! Debug log below.
Since making the diagnostics, I have since removed the router line, and left it as DNS server 1= pihole, and DNS server 2= cloudflare
My thinking was if my pihole was ever down, then cloudflare would be used as a backup. I've had it that way for years (including with the router in between as shown above, which I thought might be a potential endless loop so I removed it). This issue just started within the past week.
Your expectations are wrong. The second DNS is not a backup, it's just another option.
This will work as you expect ONLY when Pi-hole is down, but when Pi-hole is running, the devices will also use 1.1.1.1 as DNS server (at the same time). At least a few queries will bypass Pi-hole.
I am sure. Silly me, I saw that my server had 390 days of uptime with no issues and thought it was about time to update to the latest Unraid OS. After the update, I was crashing repeatedly, unable to get past ~12 hours uptime. Once I disabled pihole and kept all other containers running, I made it to 7 days uptime with no crashes. Then I read and made the change from macvlan to ipvlan, still no crashes for 3 more days, then I decided to give pihole another go and ran into this intermittent hang issue.
If you don't see any problem with running pihole on ipvlan with a custom IP on eth0, then you are right, the problem likely still lies with something they've changed with unraid networking. I just can't seem to find any way to identify by log or otherwise what is actually happening during this hang period to troubleshoot this issue.