In an attempt to see if a game server was having a hard time with the MTU settings on the default 6.41 rev of ROuterboard I reduced one interface (eth3) to 1491. Once that was done every device on the lan was experiencing random DNS issues. Some SSL sites would not resolve and others were so slow they would time out. Some non SSL sites were also affected but not all. It doesnt seem to matter what you do on devices outside of routerboard - you can, for example, put your desktop MTU at 1491 and everything is ok - but change that on the routerboard and this weird DNS issue pops up.

If this is not a bug and I simply need to read up on how this change would be expected to produce this...send me a link to read up...seems bizarre to me.

What is connected to eth3?
Were the device's DNS queries from the local DNS resolver or from an internet DNS resolver?

my desktop is connected to eth3.

I was using the DNS feature within Routerboard. I tried 8.8.8.8/8.8.4.4 and also the primary and secondary for my ISP and then all 4. These all work without issue when the MTU is at 1500 but when I downgrade to 1491 - I get random DNS weirdness. I also tried to create static DNS entries to resolve this on the nics of each device...it didnt seem to make a difference. I don't understand that at all.

This all seemed to coincide from my CAP2n going offline. Subsequent investigation showed something was odd on the POE port - it was supplying power but no ethernet connection. I had to reboot with the POE unplugged and then let the autostart POE feature power it back up. For all I know the MTU change kicked that issue off too.