configure sun cluster on nexus 9k

[sufyan.a1]

I have two sun servers connected to two nexus, where each server is connected to both nexus, when I shutdown the port the connected to MPLS, the sun cluster is going to down.

is there any special configuration for sun cluster on nexus.

regards,

[sufyan.a1]

is there anyone have idea?

[Kirk J]

Greetings.

How about a topology, and some configurations of your ports facing the Sun servers, as well as the ports going to the MPLS.

Thanks,

Kirk..

[sufyan.a1]

thank you for replay, please find attached for drawing, I check with Sun admin, and I found each server is connected with two cable to the same core, is this will effect.

thank you again

sufyan

[Kirk J]

It would helpful if you defined the following:

• What is a cluster failure?  (the cluster itself goes down, services shut down, or you can't reach the VIP ? )
• are there cluster services that rely on connectivity out your MPLS link?
• When you are 'measuring' a cluster failure, how are you communicating with the cluster from your remote workstation?  A link off the N9Ks going to an access switch, etc?
• It appears you have a VPC defined.  For the Port-channel links, what kind of channel-group config do you have?  How is the 'bond' configured on the SUN servers?

Thanks,

Kirk...

[sufyan.a1]

Hello Kirk,

- please find bleow the error on cluster server:

NIC failure detected on vnet0

of group ipmp

Apr  7 09:13:49 AMMBEEBP01 in.mpathd[335]: [ID 832587 daemon.error] Successfully failed over from

NIC vnet0 to NIC vnet1

Apr  7 09:13:56 AMMBEEBP01 in.mpathd[335]: [ID 585766 daemon.error] Cannot meet requested failure

detection time of 10000 ms on (inet vnet1) new failure detection time for group "ipmp" is 32242 ms

Apr  7 09:14:56 AMMBEEBP01 in.mpathd[335]: [ID 302819 daemon.error] Improved failure detection tim

e 16121 ms on (inet vnet1) for group "ipmp"

Apr  7 09:14:59 AMMBEEBP01 in.mpathd[335]: [ID 302819 daemon.error] Improved failure detection tim

e 10000 ms on (inet vnet1) for group "ipmp"

Apr  7 09:15:12 AMMBEEBP01 in.mpathd[335]: [ID 168056 daemon.error] All Interfaces in group ipmp h

ave failed

Apr  7 09:15:12 AMMBEEBP01 Cluster.PNM: [ID 890413 daemon.notice] ipmp: state transition from OK t

o DOWN.

vnet0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2

        inet 10.20.32.52 netmask ffffff00 broadcast 10.20.32.255

        groupname ipmp

        ether 0:14:4f:f9:74:a5

vnet1: flags=9040843<UP,BROADCAST,RUNNING,MULTICAST,DEPRECATED,IPv4,NOFAILOVER> mtu 1500 index 3

        inet 10.20.32.56 netmask ffffff00 broadcast 10.20.32.255

        groupname ipmp

        ether 0:14:4f:f8:78:70

 -are there cluster services that rely on connectivity out your MPLS link?

no it's connected to LAN Nexus.

- When you are 'measuring' a cluster failure, how are you communicating with the cluster from your remote workstation?  A link off the N9Ks going to an access switch, etc?

the connection is going down, so the users can't connect to cluster, but you can access the  N9Ks

 - It appears you have a VPC defined.  For the Port-channel links, what kind of channel-group config do you have?  How is the 'bond' configured on the SUN servers?

I have VPC between N9Ks, but no Port-channel  configured for sun cluster, I connected the Suns Servers direct.

thanks

[Rick1776]

I agree with Kirk, can answer the question until we get the entire configuration setup.

A lot is going to have to do with how the Sun severs have their NIC teaming configured and on how the port-channel/channel-group is configured.