Hi,
since some time, I have recurring issues in my network. Connections break, sites are not loading etc.
Every time that happens, CPU usage on both cores is 100%. I cannot see any single process that uses that much, but it seems it’s related to resolver.
That lasts for about 30 to 60 seconds, then it returns to normal operation
Any idea? Any more information I need to gather for diagnosis?
Model CZ11NIC23, currently on TOS 6.2.2.
You have high CPU usage in unbound but Omnias ship with kresd instead and AFAIK unbound is not officially supported in there.
I switched to unbound because it turned out that the shipped kresd is feature-incomplete.
So is there any officially supported solution that supports holding a local copy of the root zone?
Or any “unofficial” way to fix the issues with unbound?
I’m aware of no better option right now. To me the screenshot suggest that there might be a problem in the part that transports DHCP names to DNS. You could try to turn that off in reForis, at least if you don’t use them.
I do use local names a lot and for nearly every device here. They all have a static DHCP assignment. So they need to be resolvable.
You might be right, maybe DHCP lease expiration is somehow involved here. Maybe someone has an idea to troubleshoot that further?
So this seems to happen every time python3 /etc/resolver/dhcp_host_domain_ng.py is called. Any idea?
