all few weeks the dnsmasq service will not further answer.
I saw the dnsmasq process claim on whole CPU with 50 % of the system ressource.
A hangup signal of the process does not help but pressing kill button restart the process and then it works again.
Which process then is normally doing the dns service if not dnsmasq normally do it?
I have configured the omnia so that I have an internal resolving of all of my network devices.
All other should go to an external resolver.
Ok, maybe dnsmasq is used because I run the Omnia for years now with upgrading regulary only the os so knot was not used in the beginning.
I see that dnsmasq-full is installed. Only knot-libs and knot-libzscanner pkg is installed.
Is there a path known to migrate to knot? Is dnsmasq still used for the dhcp stuff?
A kresd is not running.
Maybe I can replace dnsmasq by odhcpd. Than knot should do the dns stuff.
I never recommended the use of one proc for different services so dnsmasq was never my first friend. Each service should have there own software so all can be replaced by each other.
Looks like there is a bug in dnsmasq allready known. Don’t know at the momen when it will fixed.
By the way it is not exactly a turris problem but should be fixed by an upgrade of dnsmasq after the problem has to been solved by the dnsmasq developers. It is on my watchdog list for the moment.
There exitst a problem in dnsmasq befor 2.90 which could is affected by CVE-2023-50387 and CVE-2023-50868 (extreme CPU consumption in the DNSSEC validator).
With 2.90 it is possible to break this problem.
The current version used is 2.86 so I suggest that dnsmasq should be updated.
With migration of OpenWRT 23.05 the new dnsmasq version should be ugraded so hopefully the problem will be cleared later.