[SOLVED] Draytek 2832 strange issue

[freedomit]

We have a new V2832n router and have had major issues with it. The installation site is very simple, its a serviced office so an ethernet cable comes from a Cisco switch in the comms room and gives out an private IP via DHCP. When we installed the new 2832n to replace an old Netgear router it really struggled to get an IP address via DHCP and only worked around 10% of the time. If we set the connection to static then it connected but we were getting 50-90% packet loss when pinging google. The router also would often lock up and stop responding to ping on the LAN interface so would require a reboot. I had an old Vigor 2820n in my car so i installed that and it worked perfectly first time no problem, packet loss was 0% and internet speed good. Draytek support asked us to try various things but same issue, we were also able to replicate the problem in our own workshop so the unit was replaced under RMA.

Anyway, i went back today with the new 2832n and got the exact same problem. The router locked up twice and internet was all over the place. I changed the WAN interface from auto to 100 Full Duplex and it stopped crashing but internet wasnt right. Packet loss when pinging google was 10%, web pages took ages to load and i really struggled to get a speedtest to even run. I switched back to the 2820n and everything works fine again.

The WAN settings in both the 2832 and 2820 were identical, set to always on, static IP, MTU 1500 and auto negotiate.

Has anyone got an ideas what could be causing this issue?

 

[Markverhyden]

Have you run a cable test?

Have you short cabled the new router to the switch in the comms room and tested there?

Did you do a factory reset/reflash firmware on the new box?

 

[freedomit]

Have you run a cable test?

Have you short cabled the new router to the switch in the comms room and tested there?

Did you do a factory reset/reflash firmware on the new box?

We don't have access to the comms room unfortunately, i don't even know where it is to be honest. Its a farm estate and as far as im aware they don't have a full time staff member onsite to give me access.

I was thinking maybe it was a cable issue but why would the v2820 work and the v2832 not work? The v2832 has a gigabit WAN port but setting to 100Mb should then replicate the v2820. Also why would a bad cable keep causing it to crash and go unresponsive?

Yep flashed with latest firmware and factory reset.

 

[Markverhyden]

We don't have access to the comms room unfortunately, i don't even know where it is to be honest. Its a farm estate and as far as im aware they don't have a full time staff member onsite to give me access.

I was thinking maybe it was a cable issue but why would the v2820 work and the v2832 not work? The v2832 has a gigabit WAN port but setting to 100Mb should then replicate the v2820. Also why would a bad cable keep causing it to crash and go unresponsive?

Yep flashed with latest firmware and factory reset.

Hard to say. It does sound very odd but those are typical troubleshooting steps I use in situations like that. I would tell them you need access to the comms room. And for testing the cable get your hands on a high quality certifier, not one of those blinky light things. Another thought is the Cisco switch in the comms room. What model is it? Short cabling to the comms room will tell you a lot. And for grins try 100 half duplex. I've had to do that a few times for networks with the types of symptoms you are describing.

 

[freedomit]

Hard to say. It does sound very odd but those are typical troubleshooting steps I use in situations like that. I would tell them you need access to the comms room. And for testing the cable get your hands on a high quality certifier, not one of those blinky light things. Another thought is the Cisco switch in the comms room. What model is it? Short cabling to the comms room will tell you a lot. And for grins try 100 half duplex. I've had to do that a few times for networks with the types of symptoms you are describing.

Ok thanks i have put in a request for access to comms room, i agree plugging directly into the Cisco will rule out any cable issues so i guess that's the best place to start.

 

[TurricanII]

I would knock the MTU down to 1442 on the non-working Draytek. What is the MTU on the working older Draytek?

When daisy chaining routers a few bytes can be added to packets at each hop. If you are sending bigger packets from the LAN than any upstream router is allowed to send then you end up with massive re-transmissions, high CPU use, intermittent connectivity etc. etc.

 

[freedomit]

I would knock the MTU down to 1442 on the non-working Draytek. What is the MTU on the working older Draytek?

When daisy chaining routers a few bytes can be added to packets at each hop. If you are sending bigger packets from the LAN than any upstream router is allowed to send then you end up with massive re-transmissions, high CPU use, intermittent connectivity etc. etc.

1500 on both but another thing to try so thanks

 

[Mick]

I would knock the MTU down to 1442 on the non-working Draytek

Yes - would second that. No Drayteks I've installed have enjoyed an MTU at 1500 - 1492 seems to suit them much better, although you may have to play around a bit to get the optimum for your particular situation/model

 

[freedomit]

UPDATE....
So I gained access to the Comms room, what I assumed was a Cisco switch is actually a Netgear ProSafe M4100-26G. From looking at the port lights it looks like each port has been set to 10/100. I connected directly into the switch and tested but got the same issue, pings are dropping about 10% and web pages won't load and after a while router crashes. I set the WAN port to 10/100 which seems to stop the crashing but same issues with ping packet loss and web pages not loading. I tried MTU of 1442 and 1492 but same issues. I also noticed another tenant is using a v2830 so that must work.

My theory is that there is some kind of bug in the v2832 range as they are very new out, the first one I got even shipped with RC firmware.

I'm going to try the same setup on my home network by plugging the WAN into my crappy BT broadband home router and see if the same thing happens.

 

[seedubya]

Draytek support are usually very good when it comes to stuff like this. Have you contacted them? I would given that you've tried 2 routers with the same result.

 

[Markverhyden]

Did you try half duplex?

 

[freedomit]

Did you try half duplex?

Sorry forgot to mention...yes I did but was the same

 

[freedomit]

UPDATE 2....the home test

So I connected the Draytek WAN into the back of my BT Homehub to run some tests. Internet worked great and did a bit of work while connected. I then realised the BT LAN ports are gigabit so I ran a couple of tests with the Draytek WAN set to Auto Neg and also on 100 Full, MTU was left on 1500.

Firefox - Speedtest.net Beta Version
Gigabit = 42Mbps / 8Mbps
100FD = 22Mbps / 0Mbps

IE 11 - Speedtest.net Classic Version (Flash)
Gigabit = 45Mbps / 9Mbps
100FD = 28Mbps / Fail

So it seems the issue may lie when the WAN port is set to or connected to a 10/100 connection.

Any ideas?

 

[Markverhyden]

Something I forgot to mention, changing settings on the Netgear. I'm guessing that is provided by the ISP so you may have to get level 2 support involved.

 

[freedomit]

Something I forgot to mention, changing settings on the Netgear. I'm guessing that is provided by the ISP so you may have to get level 2 support involved.

It's looked after by another IT company but they have been helpful so far. I'm going to ask tomorrow if it would be possible to switch my port number to auto rather than 10/100 I think it's locked to.

Still doesn't explain the issues I'm having at home and issues I've had when setup in my workshop.

 

[Markverhyden]

It's looked after by another IT company but they have been helpful so far. I'm going to ask tomorrow if it would be possible to switch my port number to auto rather than 10/100 I think it's locked to.

Still doesn't explain the issues I'm having at home and issues I've had when setup in my workshop.

This stuff is simple and complex at the same time. The simple part is - it works or it does not. LOL!!!! The complex is figuring out why. Had a call at a hotel last year. They had just installed a new ISP connection. Verizon FIOS (fiber) and were having all kinds of performance problems similar to what you are describing. After hours of fiddling with everything under the sun I could touch I finally got Verizon to change the settings on the ethernet port of the ONT to 100/full from auto. Poof!!! Problem solved.

 

[TurricanII]

It shouldn't be enabled by default but check for some crazy bandwidth limit on the Draytek that might be affected by WAN port speed..

As the office is managed perhaps ask the IT function to set the Netgear port to gigabit to make it work?

+1 for Mark's suggestion - in case you did not try both then force it to 100 full and half duplex - one may work and the other may not. A mismatch of duplexity = very slow/intermittent/dropped packets

 

[freedomit]

UPDATE 3....the office test

So we have a 250Mbps connection with Virgin Media Business, so i repeated my testing but plugging the Draytek WAN port into the Virgin router and running speedtest.

When Draytek WAN set to auto = 211Mbps / 14Mbps

When Draytek WAN set to 100FD and 100HD = 0.6Mbps / Fail
Also web pages take 1-2 minutes to load

So this is the third environment now where having the Draytek WAN either set to 10/100 or connected to a 10/100 router/switch the internet speed is unusably slow.

Next stepcall Draytek Support...watch this space

 

[freedomit]

UPDATE 4....SEG/Draytek Support

So ran through my results with support and they suggested setting the MTU to 1400 which i did. Same issue though, when set to AutoNeg and running at gigabit it was fine, when set to 100FD i was getting 3Mbps. The guy said it will work when set to AutoNeg and connected to a 10/100 connection, so i dug out a 10/100 switch and put it between the Virgin Router and the Draytek and guess what it kinda worked. Download was 70Mbps and web pages load fine, but upload is very low at 0.5Mbps. Im waiting for the guy to come back from lunch to see if we can improve the upload speed.

If setting the MTU to 1400 is the answer then i have questions...

1) Why does the v2820 work with MTU of 1500 fine?
2) Why doesnt the 1400 MTU work when the port is set to 100FD and connected to a gigabit switch?

 

[seedubya]

Really looks like buggy firmware.