It's pretty dumb to point both at Google and OpenDNS since they can sometimes give different answers for the same query (because OpenDNS intercepts various stuff and Google doesn't).

However that's probably not the cause of your problems.

I'm seeing what I think is the same issue. Internet will suddenly drop out, then I reboot router and it's fine.

Here's what I'm seeing:-

At some point in the past, Virgin was having some DNS issue so I setup the DHCP on the server to give out Google's DNS addresses to the client's. That resolved the issue at the time. At some point I updated the firmware to 3.3.5.2 and since then issues have been re-occurring around DNS.

What I am seeing is a bit odd - under Online Status page, the WAN2 connection DNS setting was showing up as the Google DNS entries I had put in under the DHCP config. A reboot would fix this and it would revert to Virgin's DNS servers (194.168.4.100 and 194.168.8.100)

I have double checked under the WAN2 config to make sure I haven't done something stupid like set up the DNS in there to point at Google, but it's definitely just blank, so will take the values give to it from the cable modem.

I have done a firmware upgrade to 3.3.7.3 in the hope it would fix this behaviour but it doesn't appear to have. At the moment, I've done a full factory reset and set everything up from scratch. I was rather discouraged to see that the Google DNS servers have put in yet another appearance in the DNS under the online status - and remember, this is after a factory reset! The DHCP config has 192.168.1.1 (the router's IP) as the DNS setting in there.

I'll try and grab a screenshot as evidence if/when it happens again.

Its sad to hear Draytek have not yet released an official fix for DNS packet drops.

To be clear, my issue was that the router would intermittently drop DNS query packets - and this was any DNS query packet. Queries made to the routers DNS proxy itself, and queries generated by LAN PC's that targeted external DNS servers directly. Run DNS Bench tool - it is very good at revealing the issue. With this issue present it looks like you internet connection is constantly dropping out even though its fine.

I did in the end get some beta firmware from Draytek that fixed the issue but Ive not seen it officially released yet or the fix on any release notes. If anyone wants to try this firmware you can get it from www nysc co uk downloads Draytek2930DNSFixv2930_0001.zip

Sorry for the cryptic link.

I dont know why you are getting issues with old DNS server addresses showing up. Thats a bit strange. Ive happily flogged off my 2930's and all the Draytek routers that customers had because I was sick of stuff like this and support didnt believe me about the DNS issue - I had to scream real loud. Ive gone Mikrotik (RB450's in most cases) and never looked back.

One thing I do remember though with 2930 is that if you use 2x WAN you have no control over which DNS servers from the 2 WANs gets used. It seems that the WAN that most recently got an address wins - which isnt terribly useful. I would fix the DNS server settings if I was you to ones other than your ISP's.

I have recently bought a 2920 and installed it in a small serviced-office environment with two ADSL2+ on the WAN side. After tearing my hair out over erratic problems with the Interent appearing to drop, VOIP phones deregistering, VOIP calls dropping, VOIP phones not dialling out properly when clearly on the network, QoS not behaving logically, load-balancing not working, and WAN1 / WAN2 coming in and out of operation on a whim, it occurred to me something seemed to be up with the DNS on our LAN side and that packets seemed to be getting lost in the otherwise straight forward network. I reduced everything to one LAN switch, rebooted the modems, rebooted the 2920 and rebooted the switch and still the problem persists erratically. Discovering this thread, it seems that this is a known issue that Draytek have not solved for multiple bits of kit, and renders the hardware not fit for purpose. I have wasted about 5 man-days on this problem and have a number of irate customers in our facility.

I have noted that if you switch load-balancing or traffic from one WAN to the other WAN, within the software control panel once there is a freeze, it seems to relieve the problem temporarily. This can be seen by pinging a WAN side IP address and then switching traffic to the other WAN. But there appears to be a more subtle routing issue, as even WAN1 and WAN2 interface addresses cannot be pinged at times and the internet settings of the two WAN ports are supposed to know these addresses explicitly!

Anyone know of a 2920 firmware fix please?

there is a beta here to try guys if you got time but not sure what is fixed
ftp://ftp.draytek.pl/Seria_2920/Firmware/beta/

Thanks for the firmware link. Have uploaded and will test over the next few days. However, I am still seeing problems with at least one VOIP phone after the update - it is on the LAN, registered at the VOIP provider, but wont connect an outgoing call which may still be symptomatic of lost packets. Or it may be a red herring! Will see how the rest of the devices behave...