Cluster Resource

[mbaisdon]

Having a problem with resources being not available, however if you look in Cluster Admin all resources are online and all looks good. If I fail the cluster group to the other node connections can be made to the resources. If I fail it back, connections are available for about 5 minutes then they loose connection to the resources. Anyone seen this?

 

[xmsre]

Is it a dell server built with server assist 8.0?

 

[mbaisdon]

No IBM Servers connected to an IBM FastT SAN.

 

[flaurijssens]

Which kind of resources do you mean?

 

[mbaisdon]

File resources

 

[flaurijssens]

So, it happens after you perform a fail-over from node A to node B. For a while connections to node B are available, then they are dropped.

What happens when you reboot node A after the fail-over?

If that does the trick, your network card (Broadcom?) setup might need some revision. See thread737-528154

 

[xmsre]

I was thinking the broadcom NIC also. The driver version for Dell that comes with server assist 8 does that, the version with server assist 7.5 does not.

 

[mbaisdon]

Yes Broadcom Gig NICs here. I will see if there is a new version of the driver and see if that fixes it. Thanks for your inputs all.

 

[xmsre]

the one dated 5/23/2003 works fine.

 

[mbaisdon]

Well updated NIC drivers and still same thing, any other ideas. You can still connect through the server name, just the cluster alias seems to get lost. And the other node seems to be working fine, no problems.

 

[flaurijssens]

Do you use multiple adapters teamed together? For some odd reason, newer drivers did not work for me last time I had to solve this issue (in fact, the new drivers did not work at all for these particular Broadcom fiber cards).

I ended up breaking the team and disabling all but one fiber NIC, thus leaving one Broadcom Gigabit fiber NIC for LAN access and one 100base-T for the heartbeat signal, in each node.

That did the trick.