I have a 4006 chassis connected to 2 3560Gs via two port-channels. Those two 3560's are connected to each other via another port channel. Periodically - I get the following log entries - indicating that port-channel 1 on 3560#1 is being blocked because of <see below>. At the same time, I get port channel 2 on 3560#2 with a similar logg entry. Then the next 2 messages state that it is being blocked, then it restores. I don't understand where the BPDU that it refers to is coming from... my simplified topology is after the log entries... all switches are configured for STP, and the 4006 is root with a priority of 1 (it wins).
Show logging on 3560 # 1
004633: .Nov 18 11:48:30: %SPANTREE-7-RECV_1Q_NON_TRUNK: Received 802.1Q BPDU on non trunk Port-channel1 VLAN1.
004634: .Nov 18 11:48:30: %SPANTREE-7-BLOCK_PORT_TYPE: Blocking Port-channel1 on VLAN0001. Inconsistent port type.
004635: .Nov 18 11:48:45: %SPANTREE-2-UNBLOCK_CONSIST_PORT: Unblocking Port-channel1 on VLAN0001. Port consistency restored.
I'd list the entire configuration -- but it is pretty straight forward and representative of the above. When this happens we think users are getting disconnected if they have an active SMB or IP session to either server.
I know this is fairly vague -- but tell me what else I need to explain?
If the uplinks from the 4000 to 3560s are trunked, can you check to make sure you are have the same vlans on each side of the connection including the etherchannels, and also same encapsulation an all of them?
Looking at my configurations with your suggestions I think I found the following. The port-group (etherchannels) on my 4006 are in 802.1Q trunking mode, and the Port-Channel(s) on my 3560s are not. Could you look at my attached file and see if this is obvious?
I see that on switch 3560-1 you have po3 trunked, but not po1 and the same for 3560-2 po3 is trunked but not po2. If you only have one vlan, then you can turn off dot.1q an all the switches. If you have more then one vlan, then you need to make all your portchannels dot.1q. Keep it consistent.
I understand keeping it consistent. But I don't understand why it is happening. I thought that I had set ports 5/1-6 NOT to trunk by virtue of not turning off trunking in the modle config on the 4006.. But in the default config I see the following lines for 5/1-6. This says that
trunking is auto
dot1q. In the non-default configuration
I do not call out ports 5/1-6 to trunk, as I do in ports 1/1 and 1/2. Does this then declare that they are trying to setup trunking (because of the default command), but my etherchannel on the 3560 side is not set to trunk? Maybe this is obvious but at this point I want to ensure I fully understand the implications. Since all of our traffic on the core side is on Vlan 1, I didn't think I needed to trunk this link, just etherchannel.
Your response is anticipated!
Config items mentioned above are now listed...
Default Configuration item:
set trunk 5/[1-6] auto dot1q 1-1005 (all six ports have the same line)
Non-default configuration, I DO NOT set these ports to trunk, as with other ports...
Star C> (enable) show runn 5
This command shows non-default configurations only.
Use 'show config all' to show both default and non-default configurations.
# ***** NON-DEFAULT CONFIGURATION *****
#time: Wed Nov 24 2010, 10:02:11 PST
#module 5 : 6-port 1000BaseX Ethernet
set port name 5/1 Channel to b52vm1
set port name 5/2 Channel to b52vm1
set port name 5/3 Channel to b52vm1
set port name 5/4 Channel to b52vm2
set port name 5/5 Channel to b52vm2
set udld enable 5/2-4
set port channel 5/1-5 mode on
Star C> (enable)
------------ there is no line as below with 1/1-2 ------------
We are pleased to announce availability of Beta software for 16.6.3. 16.6.3 will be the second rebuild on the 16.6 release train targeted towards Catalyst 9500/9400/9300/3850/3650 switching platforms. We are looking for early feedback from custome...