I’m running v6.42.6 on a CCR1009-8G-1S-1S+ at a large event.
Today under reasonable load (10k pps / 1Gbps WAN routing for ~1k users on ~20 vlans) i’ve experienced the Mikrotik DNS service stop responding to requests.
Clearing the cache, or changing the upstream DNS IP’s (or possibly even just clicking “apply”) seems to bring it back to life. This is making me wonder if this is a bug.. and if anyone else is having this issue.
I’ve been able to find a few posts online of the same symptoms, but they’re years old!
Here’s my DNS config.
/ip dns
set allow-remote-requests=yes cache-max-ttl=2h cache-size=8192KiB max-concurrent-queries=200 max-concurrent-tcp-sessions=40 \
servers=9.9.9.9,149.112.112.112
Hey, I’m also having the same issue and have also reached out to the Mikrotik support and their latest update to me is :
“Unfortunately we still have not managed to reproduce such problem, but we are actively trying to reproduce it in order to find and resolve the problem.”
Probably it would help Mikrotik if anybody suffering from same symptoms created a supout.rif file using command /system sup-output name=supout.rifbefore restarting DNS service. And then obviously send the said file to Mikrotik’s support. Not everybody (probably including Mikrotik) has 1k heavy users using DNS service from RB itself, probably in normal circumstances it’s fine to run DNS service on some dedicated or virtual server.
Hey mkx
I’ve already sent the details(configs, logs, other observations) to mikrotik, why would you assume that its not done ?
The only thing is that the issue is not been able to be replicated yet in their lab, probably
So.. I found the problem and it’s embaressingly simple. My router was out of disk space. The max cache size was set to 8MB but the disk only had 2.5MB remaining. Once the DNS cache reached that size, the service stopped responding entirely (that’s including cached responses) - which is not the expected behaviour but I guess makes sense.
The expected behaviour would be for a warning or at least a log entry when a full disk occurs, and for the remaining disk space to become the max cache size. DNS should not stop responding with no warning.
In my case the disk was full of package files from previous updates. I imagine some others with this issue may have disks full of log files etc. Hope this helps someone, and a fix can be made in a future release.