CIFS server causing loop and generating errors

Thanks Dynamox.

Depends on your switch – please ask your switch vendor or networking admins

You absolutely HAVE to use the same trunking on both the Celerra AND the switch

Otherwise it wont work – basic networking

Rainer, thanks for the quick response. I will check with the Network admins if it can be done on the fly or an offline change. Btw, the DM's are in Active/Passive setup.

Port-channel can be setup on the fly

it's only disruptive when you are taking existing and working network connections and configure them into port-channel. Sounds like your connections are not working right now anyway ..so can't harm anything that's not already broken.

Have you looked at ip reflect?

There is is a parameter "facility ip param reflect" that can disable ir (no reboot required)

Claude

Dynamox is correct; it is only disruptive to the ports you are applying to a trunk.  It is not disruptive to the entire switch.  This is the correct path to take; any other path to correct this will result in a much longer outage.  For example, you can remove the trunk on the Celerra side, but it is quite disruptive, more so than making a trunking change on the switch.

Also, remember to ensure that the trunking protocols match on both the switch and the Data Mover.  You said that you have a "port channel" configured.  Is it simply a port channel or is it flagged as LACP?  A simple port channel on Celerra is actually the Cisco EtherChannel trunking protocol.  To use on a Nexus switch, you would need to enable LACP (802.3ad) trunking, which is an IEEE standard.

Finally, I would not consider disabling IP Reflect.  IP Reflect simplifies management of the Celerra by sending the reply packet out of the same interface in which the request came. (which is why you are getting loop errors).  However, without this in place, your routing table must be set up correctly.  If it is not, you will cause great disruption to everything else on that server.

I would suggest making the port channel change to the switch, and I would also suggest doing this in an offline window or in a low I/O period, to minimize disruption.

-bill