Died this morning and needed a reboot. Since I needed to login for work had to reboot the router so couldn't pin down the time it went tits up. Sent the log files to draytek.

Died again this morning one pc only access to ipv4 sorts but this time a release and renew fixed it.
However when I tried to connect to work using a mac and a pc they had ipv4 issues and vpn issues so a reboot was needed.
Two other machines that run 24hrs seemed to be fine though in regards to internet access which wasnt the case in the past. So I suspect the router was at the time getting into a broken state.

qwaz1 are your issues just hitting ipv4 or both ipv4 and 6?

Hey ho! I thought I join you guys after all lol

OK, set the 3910 up yesterday afternoon. For a piece of supposedly near enterprise grade kit, it feels very lacking and unstable. I am scared to touch it or to give it too much work in regard to load balancing and QOS. It was running fine over night, but the problems started this morning. Here a few impressions:

1.) Transferring VPN settings from the 2862 via normal export was sadly not straight forward (unlike what the Draytek website said) and quite a few settings had to be changed over. But I managed to get it back up running after a couple of hours. This is a single Lan-to-Lan IPSEC VPN.

2.) I like the new dashboard which is in line with their newer switches (have a few of their large POE switches running in two locations), but it seems to be slow to update (the port states on the dashboard seem to be very slow to change and sometimes don't show up at all).

3.) This now runs three WAN connections via static IPs: 300Mbit, 73mbit and 20mbit (not part of the load balancing pool) - Load balancing is working fine and it is delivering better throughput than the 2862 did with the two main connections. CPU load is showing at around 6-8% when under "full load", while the 2862 was maxing out around 90%.

4.) This morning it started. Our Wifi 6 AP is connected via the 2.5G port but lost connection today, reboot did not help, but when I switched it over to a 1G port, it worked again. So the 2.5G ports seem to be a bit wonky. They synced correctly last night, so I will revisit. This was one of the major selling points since the Wifi 6 AP connects quite a few machines to our upstairs and the limit to 1G when connected to the Vigor 2862 was a bit of a bottleneck, when there was heavy load on the wifi and wan at the same time.

5.) Shortly after that some machines suddenly started to show a huge number of IPv6 addresses. I have no idea what caused it and if it was to be expected, but after disabling it on the router, they reverted to normal (one ipv6 per interface). I am no IT expert and have not got a clue about IPv6, so take it with a pinch of salt.

It looks like I hit the maxium allowed characters lol, number 6 with conclusion follows as soon as the spam protection let's me post again ...

here the rest:

6.) An hour later we had the big collapse of the whole thing, basically disabling the router. All WANS going down, unable to log in to the web interface. Pings not getting through. The only way to access the interface was to unplug all WAN connections and connect with a direct Ethernet connection. Reconnecting the WAN connections did not do anything, the router just showed all connections as disconnected and ports as down (even though they showed as synced and active on the front of the machine). CPU showed as 1%, memory showed as 1% (so probably false). A simple reboot did not solve it, still all connections and ports showing as inactive. It had to be completely power cycled with all WAN connections removed and then added back one by one.

The root cause seems to have been a Sky Q box. When googling it, they seem to cause havoc across a range of installations (including Unify etc.), causing broadcast storms etc. We had issue with the box in the past, but it never killed off the 2862 as it did with the 3910. To make sure it wasn't a loop via the switch, I connected the box directly to the 3910 and it disabled it again within 5 minutes. So box is now completely disconnected and the router is running since, so I will observe this and report back, but if you run a Sky Q box, it might be worth investigating.

Overall it's a bit disheartening, that this can cause so much instability. The firmware seems to be running unstable overall, it does not feel solid at all and I have the feeling the whole thing will crash any minute, if I demand too much of it (which is sad, because it is only running 400 mbit combined WAN connection, one VPN and maybe 20-30 client machines).

Netjet1980 - Cant say we didnt warn you!!
You are brave or foolish!

It does appear the 2.5 ports are very buggy, I only use mine as 1Gbe WAN so cant comment apart from they are fine for that purpose.
The two SFP ports for me are running faultlessly, port 2 is my 10Gbe LAN feed and port 1 is set as a Lan port mirror with a RJ45 to SFP+ adaptor, this feeds some monitoring software.

Dont use QOS, no need. Load balancing is little bit buggy, it does work sending some ports to different connections. A while ago I tried to add a Domain based route policy and it died a death, had to reset and reload config.

They are a bit slow with firmware updates, still waiting for USB to be available for Syslog and config auto backups hopefully.

Edit: I see what you said about Sky Q - I can confirm ive seen those exact same symptoms, I did wonder about the broadcast storm thing and I amended my switches to try and mitigate that - didnt think about the Sky Q boxes, will certainly bare that in mind if there is a next time - thanks.