Switch Cluster/Clustering 2960s

Scott Cannon · ‎09-19-2010

Hi Guys,

I have 4x WS-C2960-24TT-L switches all connected via single uplink to another C2960-24TT-L switch (hence forth called the Master Switch) and share a common VLAN (VLAN 1). All switches are communicating over CDP succesfully.

The master switch is configured as a cluster commander for a cluster called Cluster1. Each of the remaining C2960-24TT-L's (hence forth called Slave A, B, C and D) are configured with with the master swithces VLAN 1 interface MAC address, VLAN number and member number using the following commands:

Slave_A(config)#cluster commander-address 0000.0000.0000.0001 member 1 name Cluster1 vlan 1

Slave_B(config)#cluster commander-address 0000.0000.0000.0001 member 2 name Cluster1 vlan 1

Slave_C(config)#cluster commander-address 0000.0000.0000.0001 member 3 name Cluster1 vlan 1

Slave_D(config)#cluster commander-address 0000.0000.0000.0001 member 4 name Cluster1 vlan 1

All switches have their own IP address on the VLAN 1 interface and all are communicating between each other via ICMP

From the master switch I can access Slave A using the Master#rc 1 command. However if I try and access any other switch, it simply does nothing (ie. comand enters without error and I'm represented with the exec prompt as though I had successfully typed a command.

Can anyone help me ascertain what the cause of this might be?

The only thing I've done that was (in my opinion) out of the ordinary and worth mentioning pertains to the MAC addresses used on the master switch to define the slaves. When I was typing the slave switches MAC address in the master switch using the following command Master(config)#cluster member 1 mac-address 0000.0000.0000.001A I found tha the master couldnt find the switch on this MAC address. Instead, I had to change the MAC address in the above example to either 0000.0000.0000.000A or 008A. When I did this the master accepted the command.

NB: There is nothing connected to these switches, they are simply interconnected as already explained.

I've attached a diag to help depict my setup

If anyone can offer any insight into my problem I'd greatly appreciate it.

TIA
Rgds

Scott

Leo Laohoo · ‎09-19-2010

Slave_A(config)#cluster commander-address 0000.0000.0000.0001 member 1 name Cluster1 vlan 1
Slave_B(config)#cluster commander-address 0000.0000.0000.0001 member 2 name Cluster1 vlan 1
Slave_C(config)#cluster commander-address 0000.0000.0000.0001 member 3 name Cluster1 vlan 1
Slave_D(config)#cluster commander-address 0000.0000.0000.0001 member 4 name Cluster1 vlan 1

I've done clustering but I sure don't use this command. I use the combo:

1. Enable cluster

2. Find out what suitable candidates ("sh cluster candidate" and take note of the MAC addresses);

3. Use the command "cluster member mac "

Hope this helps.

Please don't forget to rate useful posts. Thanks.

Scott Cannon · ‎09-19-2010

Hi Leolaohoo,

Thanks for the prompt feedback. I wasnt aware of the show cluster candidates command. Thanks for the heads-up. Unfortunately, that command shows the MAC addresses already in use and I am applying the command (in line wiht my fake example above) Master(config)#cluster member 1 mac-address 0000.0000.0000.00x1 to the master switch to configure this.

I then apply the cluster commander-address 0000.0000.0000.0001 member 1 name Cluster1 vlan 1 command on the appropriate slave switch.

Works a treat for Slave_A, not at all for any of the others.

FYI all switches are running 12.2(55)SE

Rgds

Scott

Leo Laohoo · ‎09-20-2010

On the cluster commander, get rid of your cluster commands (command: no cluster enable ) and then try the procedure I've posted.

Hope this helps.

Scott Cannon · ‎09-20-2010

Hi Leolaohoo,

Good suggestion, unfortunately it didnt bear any fruit. Symtpoms persist.

Open to any other ideas you may have.

Rgds

Scott

Leo Laohoo · ‎09-20-2010

Can you post your "sh run"?

And can you also confirm, since you are using VLAN 1, that all the switches have VLAN 1 enabled?

Scott Cannon · ‎09-20-2010

Please find attached Show Run for all 5 switches. Note: switch AUCINFSWI01 is the cluster commander

TIA

Rgds

Scott

Leo Laohoo · ‎09-20-2010

Thanks for the heads-up. Unfortunately, that command shows the MAC addresses already in use and I am applying the command (in line wiht my fake example above) Master(config)#cluster member 1 mac-address 0000.0000.0000.00x1 to the master switch to configure this.

I then apply the cluster commander-address 0000.0000.0000.0001 member 1 name Cluster1 vlan 1 command on the appropriate slave switch.

I don't see anything wrong with the config. But the statement above is not right. You don't need to apply anything to the slaves ... well not YOU, per se. Once you've entered the "cluster member 1 mac " command the commander will contact the slave and enter the appropriate command.

Scott Cannon · ‎09-20-2010

Its a wierd one eh....

I've removed the config from both master and slave and re-applied to only the master. Still no joy.

What I cant undersand is why one works and the others dont. Open to any suggestions at this point in time.

Cheers

Scott

ACOA-CISCO · ‎01-04-2012

I have a similar problem and I have not been able to fix it... but I have been able to identify a relationship with vty "transport input ssh" and rcommand. When I enable the transport input open to telnet , the rcommand works. This is not practicle for my environment as we moved over to SSH to support security recommendations from the security officer.

Does anyone know how to support rcommand with telnet disabled and SSH enabled.

Message was edited by: ACOA CISCO