Please follow the below template, it will help us to help you!
Expected Behaviour:
PiHole works perfectly -- as it has for many months
Actual Behaviour:
"Lost Connection to API" -- then spurious history displaying
Debug Token:
x10yet3bas
This morning I found PiHole inoperative "lost connection to API." With another installation I found there were issues using DNSSEC (see ticket from 23 days ago and one prior). So, I disabled DNSSEC -- the solution that worked elsewhere. However this time the problem was not solved. PiHole went off-line again. When I rebooted it it showed the following screen:
Something made these queries, and the sheer volume appears to have overloaded your Pi-Hole. Let's figure out what the millions of queries are looking for. From the Pi terminal, run these commands and post the output here.
Hi @jfb. Thanks. I found it. I later created a .PDF file showing the offending client but I can't upload that document type here. The commands you listed don't show the issue -- I suspect because of the lapse in time between the event and just now when I issued the commands. The problem was a Tycon module (https://tyconsystems.com/products/tycon-solar/all/item/129-tpdin-monitor-web2) trying to resolve the address of a NTP server -- pool.ntp.org.
I took the client off-line and rebooted the PiHole. Interestingly, the PiHole went off-line again. I then took the same action as I did with another PiHole had a similar problem -- changed from DNSSEC. That seemed to "fix" the issue. So, I must wonder if the DNS flood and the DNSSEC issues are entirely separate. I don't know.
I have three PiHoles and now two of them appear to have had the "DNSSEC issue" (pointed to "Quad 9.)"
Not sure where to go now but I'll note that all seems OK now. Very curious!
Unfortunately, FTL can crash if the system runs entirely out of memory.
With over 2 million queries, this is quite likely if you have little RAM (like on a Raspberry Pi). There isn't all that much we could do about it, maybe we could delete older queries sooner. I will check if there are any feasible options available.
But I'm wondering if the volume/memory issue would explain why FTL died again after the Pi was rebooted. And, why it seems OK again after DNSSEC was no longer selected.
This happens because FTL imports all the history back from the database on startup. As if it was never shut down. I cannot really explain why it didn't happen after you switched off DNSSEC, but it may just be that enough queries are now outside the imported time frame such that the memory is not used up any more. It may be completely independent of DNSSEC being enabled or not.
