This post is more than 5 years old
20 Posts
0
3676
Settings for NX4 to communicate with Cisco switch
I am in the process of setting up a trunk between my NX4 and my 3560G switch, but I have been unsuccessful in doing so. I'm beginning to think that there is some setting or configuration that needs to be made on the NAS that will allow communication between the two. Has anyone out there ever had a situation like this one, and if so, how did you remedy it?
bstuart1
23 Posts
0
March 11th, 2010 08:00
Ah, I understand now. You need to delete the interface on CGE0 to free it up to use in the LACP. Once you create the LACP, then you will go back to the interfaces tab and assign it an IP.
Also, CGE2 and CGE3 are not usually for "failover". All 4 ports on the promary datamover are active. You can configure a FSN (fail safe network) where one CGE port is active and the other is standby, but I think you are looking for a 2 GB trunk correct?
dwightkschrute
20 Posts
0
March 11th, 2010 07:00
bstuart,
I am using LACP.
Depending on which side you are referring. The Cisco trunk was created before the ports were connected, but on the Celerra side of things, there isn't a trunk configured. I created a trunk on cge1 for the active datamover, and it didn't work. I also tried to include cge0 in the trunk, but it didn't show it as an available option, so I deleted the trunk altogether. I haven't tried reloading the switch, either.
bstuart1
23 Posts
0
March 11th, 2010 07:00
In celerra manager, have you created a link aggregation with 2 of the CGE ports? Network -> Devices ? ?
bstuart1
23 Posts
0
March 11th, 2010 07:00
If both ports were connected before configuring the trunk, it will often not come up till you down and then up the ports on the Cisco side. Are you using LACP?
dwightkschrute
20 Posts
0
March 11th, 2010 08:00
I did just remember, however, that even logged in as root, I don't have the ability to delete the interface. Not sure why that is. If it can't be done in the GUI, will it be worth it to try it using the CLI?
dwightkschrute
20 Posts
0
March 11th, 2010 08:00
Yes, that is essentially the idea. Sorry for the misunderstanding concerning cge2 - 3. As long as, if either cge port goes down on the NX4, the data will continue to flow, and vice versa on the switch, too.
bstuart1
23 Posts
0
March 11th, 2010 08:00
The difference is Active/Active versus Active/Passive and switch redundancy. Building an Etherchannel or LACP needs to be done to the same switch, unless you have them stacked. An FSN can provide you active/passive, say CGE2 active and CGE3 passive and one port runs to one switch, the other port to a second switch, providing N+1 redundancy, but at the cost of only 1 GB thoughput. The optimum config is LACP between two stacked switches. You said you had a 3600 series right? You can't stack that switch, so both CGE ports are going to run back to the same switch, giving you a single point of failure.
bstuart1
23 Posts
0
March 11th, 2010 08:00
What ever you do with CGE0 - CGE3 for the first blade, you need to replicate that for the standby blade on the switch. (for example the LACP trunk) That way if server_2 fails, when server_3 takes over, it has the same connectivity.
dwightkschrute
20 Posts
0
March 11th, 2010 08:00
That's exactly it, it doesn't give me the option to include cge0, just cge1, so I would only have 1 port available in which to do that, and that seems rather pointless, since that defeats the purpose of link aggregation. I have cge2 - 3 available, but I believe those are for failover purposes. I know that the cge0 interface was labeled as a port in the Network > Interfaces, and it doesn't give me the option to modify (delete) it. Any ideas on how I can get the cge0 to be made available to trunk?
bstuart1
23 Posts
1
March 11th, 2010 08:00
Do you have a CIFS server or something activley setup using that Interface? That would stop you from being able to delete it. if you would liek to try the command line, I believe the command would be server_ifconfig server_2 -delete
dwightkschrute
20 Posts
0
March 11th, 2010 08:00
Actually, I don't have stacked switches, just the one, and it is a 3560G. Now, my NX4 has dual blades, so does that make a difference as far as the configuration is concerned?
dwightkschrute
20 Posts
0
March 11th, 2010 13:00
Your suggestion with deleting the interface and creating a new interface using the device I created worked wonderfully. It produced favorable results, but were not 100% there yet. I can't ping the two hosts in one of the VLANs. Everything else pings as it should. This may not be anything in which you can help me.
bstuart1
23 Posts
0
March 11th, 2010 13:00
Are the hosts you can't ping on another subnet? I have found that I need to create a 'route' on the routing tab to ping items on other subnets from the EMC.
On the routing tab I add a route with the destination 0.0.0.0, I add the gateway for the network I am going to use, and a netmask of 0.0.0.0....
As far as I can tell, this is the equivalent of adding a default gateway to the data mover.
dwightkschrute
20 Posts
0
March 11th, 2010 14:00
No, the hosts aren't in another subnet. Every host is in the same network, and
only VLAN tagging is what separates the traffic. I saw the routing tab on the
Celerra Manager, and I basically read what amounts to traffic heading to
x.x.1.0 will go out interface x.x.1.40 that uses netmask 255.255.25.0 to the
default gateway x.x.1.40. I can try your suggestion, but I'm nowhere near the
console at the moment, so it will have to wait until tomorrow before I can
test.
dwightkschrute
20 Posts
0
March 12th, 2010 10:00
With or without creating the "default gateway" on the NX4, I was only able to communicate with the hosts if I removed the VLAN ID in VI Client (VMware).
I created a routing port on the switch, so that I could test the switch's ability to communicate with every node in the network, and it was successful, again, only if I removed the VLAN ID. I know how to set up VLANs and all of that, so I'm confident it isn't a switch configuration issue. Any more thoughts or ideas?