Trying to configure an LACP LAG with a GSM7352S and our Netapp FAS2020 SAN but I can only ever get one port to come up.
We have an 8 switch stack (GSM7352S). We're using 2/0/14 and 2/0/16 on the to create the LAG. When I do a 'show port-channel all' it shows that 2/0/14 is active and 2/0/16 is inactive (Port Active = false is what it acutally says).
Config is STP=disabled, Static=disabled, send trap=disabled.
Both ports are connected to the same dual port nic on controller 1 on the FAS2020. We created an LACP vif, load balancing on IP. 'vif status vif1' on the FAS2020 shows one ports status as lag_inactive.
If I create a dynamic LAG between 2 switches (1/0/1 and 1/0/2 on one switch connected to 1/0/1 and 1/0/2 on another switch, both ports come up.
Any ideas why the second port won't come up when we connect the FAS?
Your ideas/advice would be much appreciated. If you need more detailed info or screen dumps, let me know.
Did you plug in the cables and then configure the LAG or did you configure the LAG and then plug in the cables? The order matters. Spanning tree may have disabled a port if you plugged in the cables before configuring the LAG. I know you said you disabled spanning tree, but I'm presuming that was done as part of the LAG configuration, so spanning tree may have been on when the cables were plugged in if the cables were plugged in first.
Another possibility is that the port in question is actually admin'ed down. Have you checked that?
Thanks for your response. I had wondered if that was the case without actually knowing spanning-tree would be the cause - must have read it somewhere.
I had the ports admin'd down when I configured the lag thinking that might be enough. I guess spanning tree being disabled on the lag doesn't mean its disabled on the physical ports?
I will disconnect the cables and delete and recreated the LAG before I reconnect the cables but I wonder if I'm configuring the LAG correctly in the first place. Could you tell me roughly how you would configure it?
I'm guessing you.. 1. Create the port-channel in global config mode. load-balance - Would I want to use 5 or 6?
linktrap - I'm assuming this only sends snmp traps for monitoring? system - Configure port channel system priority (LAG)???
2. Go into physical interface mode for 4/0/1 addport 0/3/1 3. Go into physical interface mode for 4/0/2 addport 0/3/1
Do I need to change anything under lacp in physical interface mode? actor Configure LACP actor. admin Configure LACP admin key. collector Configure LACP collector Max delay time. partner Configure LACP partner.
Do I need to change anything under the port channel interface mode 0/3/1?
Is there anywhere/anything else I should be configuring??
Figured it out after looking at a similar post where the solution was that the SAN engineer had split the 2 controllers over 2 switches which inspired me to try both LAG members on the same switch and *Bish*Bosh*Wallup* Up they came!!!
Looks like Netapp/Netgear don't like to play nice when using cross-stack LAG. I'm sure the GSM7352S supports cross stack LACP because it allows you to configure it both in the web console and line console. Could it be that there's a config that will allow me to cross-stack LACP LAG?
Figured it out after looking at a similar post where the solution was that the SAN engineer had split the 2 controllers over 2 switches which inspired me to try both LAG members on the same switch and *Bish*Bosh*Wallup* Up they came!!!
Looks like Netapp/Netgear don't like to play nice when using cross-stack LAG. I'm sure the GSM7352S supports cross stack LACP because it allows you to configure it both in the web console and line console. Could it be that there's a config that will allow me to cross-stack LACP LAG?
Thanks again, Pat
We use LAGs that use ports from multiple stack members with our GSM7352Sv1 stack all the time and they work fine. That said, we're using the dedicated stacking modules that used to be required to stack GSM7352Sv1.
How are you stacking? Do you have GSM7352Sv1 or GSM7352Sv2?
We're using 8 GSM7352S v1's using the 24Gbps stacking modules.
What are you LAG'ing? I haven't tested it but I'm assuming cross stack LAGs between switches will work but have you tried it with a Netapp FAS2020 SAN?
I figured it was a difference in the way Netapp and Netgear chose to implement LACP but it could just be a configuration that I've missed. I'll be vlan'ing and subnetting before I turn on jumbo frames so when I've done all that, I'll come back to the cross stack LAG issue.
Chances are, we will be buying a couple of Cisco 3750-48T's for our storage/server network which I'm certain will do cross stack LAG with the Fas2020 but if you think I can do the same with a couple of GSM7352S's, I can save about £9000!!!
We've run LAGs to Apple Xserve (LACP), Dell PE2950 running CentOS (static LAGs), Supermicro "whitebox" with a Xeon 5500 series main board running OpenSolaris (LACP), and some old 3Com 100Mbps switches (static LAGs). They're all using 2 port LAGs that go to two different GSM7352Sv1 in the same stack. They all work just fine.
It should work. Maybe the Netapp is using a hinky LACP implementation?
That said, your original post didn't say anything about this being done on multiple switches. All the LAGs you referenced used both ports from the same switch. If you were doing this across multiple switches in your stack, the port numbers in the LAG would be something like 1/0/14 and 2/0/14 not 2/0/14 and 2/0/16
Are you certain you don't have other issues here? Dropping an extra $12K on Cisco gear just to solve this problem seems premature.
I still think you may have another problem affecting one of the ports you tried originally. It could be just a simple thing like one of the ports is configured just a little differently or the second cable is plugged into the wrong port. It could be a misapplied ACL. There are numerous problems it could be, but it's next to impossible to figure out without seeing the actual configuration on both ends.
What firmware version are you running on your switch stack?
Both SAN1 vif1 and SAN2 vif1 are in LACP mode, set to load balance on IP. Mtu's are default (1500). No ACLs, no VLANs.
On the switch stack we had no VLAN's, no ACL's, no subnetting, no QOS. Basically no advanced features. I had reset the stack to factory defaults and then configured the LAG's and NOTHING else. Everything is in the management VLAN(1). Mtus are default.
With both interfaces on controller 1 or controller 2, connected to different switches, only 1 port comes up. To get it to work, the ONLY config change was to change swith ports 2/0/16 and 2/0/39 on the LAGs and physically, and then both come up as expected. No changes were made on the SAN to get it to work.
I wondered if it was port security or some other config that was causing the problem but I'm a bit of a jack of all trades - master of none!! So I'm not sure.
The firmware version on the switch stack is the most recent - 8.0.3.15. The Netapp FAS2020 is running Ontap 7.3.3.
I've been messing around on creating VLAN's since then but its not saved so tonight, I will reload the stack and then send you the config working and the config not working to see if you spot anything.
I'm not sure how I would get you the config on the Netapp but I'm happy to try if you think it will be helpful?
I think I got the idea. It only took about 15 seconds of puzzling to figure out what you were trying to show. 😉
Maybe an example switch config from one of our working LACP LAGs will help. I've redacted the VLAN information, changed the names, and changed the port numbers, but the rest is exactly as we're using it.
interface 1/0/5 description 'xserve_eth2' no spanning-tree edgeport exit
interface 2/0/5 description 'xserve_eth1' no spanning-tree edgeport exit
interface lag 1 description 'xserve_bond0' lacp collector max-delay 0 exit
exit
You might have to order the commands a little differently when entering them, but that is the order it shows in our config from "sh run". This config works.
I'm not familiar with the NetApp in particular, but I can't imagine it is all that different from other networking devices. I'll have to leave that as an exercise for you. 😉
If you still can't get it to work, try something other than the NetApp with the LACP LAG. This might be peculiar to the NetApp.
Another thing you could try is a static LAG. To make a static LAG (LACP is turned off here, though not immediately apparent) try this config on the switch. I believe if you set the load-balance method, you're turning off LACP and making a static LAG. Maybe this explains why LACP isn't working for you?
