I'm
attempting to replicate here in the office and find the solution for a problem
that we are experiencing at both our data centre sites.This is the problem:
Everytime
we load the IPMI WebUI to any of our Loadbalancer Supermicro units the IPMI
crashes shortly after logging in and connectivity can only be restored by power
cycling the Loadbalancer or disconnecting/reconnecting the network.
There
are two Loadbalancers in each rack as Master/Slave.
The
IPMI ports are connected to stacked Cisco WS-C3750X-24T-L switches into a management VLAN.
As
a potential solution we have decided to have a switch just for the IPMI/RSAs
but this is currently not working as expected as there is no connectivity from
the internal network to the IPMI port even though the VLAN is correct.
The
crossed cables connecting the 3750 swtiches and 2950 switch are configured as
lacp.
Is
this a standard configuration or is there and obvious problem with how we're
set up?
There
needs to be failover in the event that one of the stacked switches fails. The
only single point of failure is the 2950, this isn't ideal but most of the
servers in the rack only have a single remote management NIC.
Below
is a diagram. If you need any more information just ask.
You
already have VLAN tagging with the "switchport mode trunk" and
"switchport trunk encasulation dot1q" commands. What you need is VLAN
103 and 104 on both the WS-C3750X-24T-S and the 2960 in the vlan database ("show
vlan" will tell you if they are there). Then you add a "interface
vlan 103" to the 3750 and give it the IP address of the default gateway of
your load balancer. And you give your 2960 an "interface vlan 104"
and set the IP address to an address of the subnet that you will use for
management.
When
in doubt please post configs without passwords/keys.
没有评论:
发表评论