We
have 2 core switches 4506 & 6509 with 4 cisco WS-C3750X-48T-L stacks (8 switches per
stack) across all 4 floors. we have an etherchannel of 2 * 1 GB ports between
the cores. There is general slowness across the network and i cant really see
any direct cause of it, We are also getting these alerts on most of our edge
switches.
We
need to add another stack of cisco 3750 switches for new 5th floor and i was
wondering if our current setup will be able to cope with it.
5d00h:
%CMP-CLUSTER_MEMBER_1-5-NBR_UPD_SIZE_TOO_BIG: Number of neighbors in neig
hbor
update is 128, maximum number of neighbors allowed in neighbor update is 10
Error
Message CMP-5-NBR_UPD_SIZE_TOO_BIG:
Number of neighbors in neighbor update is [int], maximum number of neighbors
allowed in neighbor update is [int].
Note This message applies to the Catalyst
3750-E and 3560-E switches.
Explanation The number of cluster neighbors in the
clustering neighbor update packet exceeds the number of neighbors supported by
the clustering module. The first [int] is the new number of neighbors, and the
second [int] the maximum number of neighbors.
Recommended
Action No action is required."
How
are the stacks connected to the core switches? Are all switches configured the
same, and are they all getting the messages?
Do
you have the "cluster enable" command configured? Please post the
config of a sample switch that's getting the message.
I
have no idea whether the channel is a bottleneck, it depends on how much
traffic is running between them. You can also add ports to the channel which is
cheaper than moving to 10G and will give you a quick answer on whether that's
an issue.
The
mask on the server VLAN doesn't matter if the number of devices isn't
outrageous.
I
do see that some of the switches are not part of the stack config. I don't know
if they're gone or just unstacked, but the latter might cause issues.
I
would look at spanning tree. Take the time to determine exactly how everything
is connected, where the spanning tree root is, and what links are blocked. Draw
out an accurate topology and then mark up where blocking occurs, then post here
if you like. It could be that a poor spanning tree design is causing problems.
Regarding
the messages, my understanding is that you can safely remove the "cluster
enable" command from your switches, and that is what causing the messages.
Here is the Cisco info about this command:
Use
the cluster enable global configuration command on a command-capable switch to
enable it as the cluster command switch, assign a cluster name, and to optionally
assign a member number to it. Use the no form of the command to remove all
members and to make the cluster command switch a candidate switch.
cluster
enable name [command-switch-member-number]
no
cluster enable
Syntax
Description
name
Name
of the cluster up to 31 characters. Valid characters include only
alphanumerics, dashes, and underscores.
command-switch-member-number
(Optional)
Assign a member number to the cluster command switch of the cluster. The range
is 0 to 15.
Defaults
The
switch is not a cluster command switch.
No
cluster name is defined.
The
member number is 0 when the switch is the cluster command switch.
Command
Modes
Global
configuration
Command
History
Release
Modification
12.1(11)AX
This
command was introduced.
Usage
Guidelines
Enter
this command on any command-capable switch that is not part of any cluster.
This command fails if a device is already configured as a member of the
cluster.
You
must name the cluster when you enable the cluster command switch. If the switch
is already configured as the cluster command switch, this command changes the
cluster name if it is different from the previous cluster name.
Examples
This
example shows how to enable the cluster command switch, name the cluster, and
set the cluster command switch member number to 4.
Switch(config)# WS-C3750X-48T-S cluster enable Engineering-IDF4 4
