2 x 2700 Series VLAN Trunk

Question

I'm having a very intresting problem with 2 2716's   I have the two switches tied together via port 1 and I want to span VLAN's accross the 2 switches, basically 1 port per VLAN on each switch.  I have configured port 1 on both switches as PVID 4095 and Set ADMIT TAG ONLY.  I then created VLANs 102 - 108 and set the PVID for port 2 to 102 on both switches, port 3 to 103 on both, etc...   I then configured each vlan as tagged on port 1 and untagged on its corresponding access port:  IE VLAN 102 is tagged on Port 1 and Untagged on port 2 on both switches.   As a test, I have a PE Server connected to port 2 on switch 1 and I connect my laptop to port 2 on switch 2.   I can ping accross the link no problem, but anything more sophisticated than that locks up immediatley, file sharing, RDP etc.   With no VLAN's configured, this works fine.  I can plug my laptop into either switch when the two machines communicate normally.   Any Ideas???

TOdoubleD · Answer

By locking up I mean that the connection establishes, but RDP for instance just hangs and I have to terminate the application.  Network browsing will list folders, but I can access or transfer any data.   I have not tried forcing the speed and duplex yet.  The Server is GigE, but the host I was trying to connect from is 10/100.  I will try that and post the result.

DELL-Cuong N. · Answer

The speed / duplex setting that we are concerned about is between the host and the switch port to which it is connected and between the server and the switch port to which it is connected. It doesn't really matter if the client is operating at 10/100 and the server is at Gig. It only matter how the speed/duplex are negotiated between the host/server and the switch.

If you did nothing then likely the switch is currently setup to auto-negotiate and is setup to "advertise" a number of options to the host/server NICs. If the NICs on the host or server are setup to be fixed speed/duplex then likely you will get duplex or speed mismatch problem if the switch port is doing auto-negotiation. If you setup both switch and NICs on host and server to auto-negotiation then there is one more thing to look into.

It has been found in the past that some NICs do not properly implement auto-negotiation and will incorrectly negotiate the speed or duplex if the NIC only supports speed of 10/100 and the PowerConnect switch advertise Gigabit. So you may try either forcing speed/duplex or setup the switch auto-negotiation so that it only offer speed that the NIC is capable of supporting.

Cuong.

DELL-Cuong N. · Answer

Couple of questions for you:

When you say "locks up" what exactly is the symtoms you are observing? Is it just the session/application itself that locks up or the switch? Meaning that let's say the file transfer fails then you stop the file transfer and starts a ping then does the ping still work without having to do anything with the switch? I'm trying to figure out whether its just a bandwidth / performance thing.
So if you have everything in default mode without any VLAN then the same test where you connect a host on one switch on port 2 and another host on the other switch on port 2 will work? Meaning you do not observe the "locks up" or performance issue for the exact same test with the exact same physical setup without VLAN configured?
Can you verify for me what the port configuration is for the switch onto which you connected your hosts? What I'm looking for is whether you had auto-negotiation enabled or fixed speed/duplex on the port. I just wanted to make sure that you don't have speed/duplex mismatch on the PC and switch. If a port config mismatch occurs then you may observe performance problems even if things work for awhile.

Cuong.

TOdoubleD · Answer

Ok,   I ran more tests.   with the 2 2716, I'm using ports 1-8 with VLAN as I described in my previous post. I have left ports 9-16 in the default config VLAN1 PVID 1.   I played with the port settings and the problem still exists.  With the VLAN's involved, any connection I try and establish across the trunk fails.  It establishes a connection and then locks up like there is packet loss or something.   To review, I'm using port 1 and the trunk between the 2 2716's and right now the server is connected to port 4 of switch 1 and my host is connected to port 4 of switch 2.  Port 4 is configured as PVID 104 and VLAN 104 is configured to be untagged on port 4 and tagged on port 1 on both switches.  Port 1's PVID is 4095 and set to admit tagged only.     If I take the connection between the switches and move the interconnect from port 1 to port 9 on each switch, now tying the default VLAN together, and move the server and host to port 10 on their respective switches, the connection goes through with no problem whatsoever.  I should also not, that I didn;t tough the auto-neg/Speed-duplex on any of these port, so i think this problem is related to vlan fuctionality and not port mismatch.   DD

TOdoubleD · Answer

I'm using both NetBIOS and RDP to test.   RDP establishes and I get the Background, and then thats it.   NetBIOS is much of the same, I can browse, but its painfully slow, and then when i finally click on a folder to get access to files, explorer hangs and thats that.   Initially this conection was being used for Veritas Backup Exec traffic, and it was much of the same problem.  Application hangs when I go to access a remote resource. I though it might just be Backup Exec at first, but they I realizes it was happening no matter what app was involved.   DD

DELL-Cuong N. · Answer

Hmm...OK, I will try to reproduce the problem in my test network. I will test with FTP between two systems over the same network setup you have. If I can move large amount of information between these two 27xx using the same config you have then perhaps we need to look at your specific applications? Is there something else I should test for aside from ping and sending large data between the two systems?

Cuong.

DELL-Cuong N. · Answer

I have reproduced the problem and although I'm not 100% sure why it happens or exactly what the condition is that would cause this problem, I have discovered that any packets exceeding 1468 bytes across a trunk port is being mis-catagorized as a "jumbo frame" which cause the switch to drop the packet. So for now you might try enabling jumbo frame support on the 27xx. I only tested with ping (try pinging large packet greater then 1468 bytes and you see the same problem) and sending large files through FTP so I can't be 100% sure it works with all your applications but you might try this to see if it works for you. Let me know either way though.

In the meanwhile I will consult with other PowerConnect engineers to see if anyone has any idea why this is occuring or perhaps have heard of this issue already. When I find out something more specific I will let you know.

Cuong.

DELL-Cuong N. · Answer

The 2708 does not support jumbo frame but then I don't know that the 2708 have this problem either. So I think we need to test with it before we can say for sure whether the problem occurs on the 2708. I'm not in my lab now but when I get back there I will try it.

Cuong.

TOdoubleD · Answer

Enabling Jumbo Frame Support has fixed the problem, or at least is a workaround for now.   BTW, does the 2708 support Jumbo Frames????  I have another switch I'm going to tie into this guy which will be a 2708.   DD

DELL-Cuong N. · Answer

We have just tested the 2708.  It does not have this bug.  It does not support Jumbo frame but also does not have this problem either, so it should be fine for you to put the 2708 into the mix without any problem.   Cuong.

TOdoubleD · Answer

Thanks alot.   This board is very helpful and quick.   DD

Networking General

2 x 2700 Series VLAN Trunk

Was this post helpful?