A second unbound process started on its own, broke DNS resolver service
-
At approximately 9:20 AM EST, a second unbound process started running on my pfSense firewall. When this happened, they both seemingly began trying to open port 53, 853, etc. and clients could not resolve DNS queries.
This did not occur during an expected filter reload, pfBlockerNG cron or update job, and was not triggered by any particular action or scheduled event that I can identify. I do not have any DHCP leases being registered in DNS.
Restarting unbound through the GUI did not help. I ended up having to SSH into the firewall and manually kill -9 the older unbound process, after which the second process was able to resume servicing DNS queries.
I'm not sure if there are any relevant logs I can gather -- I'm finding nothing interesting in the system log or unbound log that points to why this happened. But I guess if anyone else has encountered this or if developers are interested in more info, I'm happy to look for any other details that might have been logged.
Edit: unbound just crashed again and I found this in the log:
error: pythonmod: Exception occurred in function operate, event: module_event_moddoneNot sure why this is surfacing now. I've been running the python module for over a year with 22.05, and have been running 23.01 now for several days without issue.
Edit 2: Found the issue. unbound is filling /var to capacity, at which point it starts failing to operate properly.