cancel
Showing results for 
Search instead for 
Did you mean: 

creating mlag and switching dowsnstream devices

creating mlag and switching dowsnstream devices

wsalkos
New Contributor II

Hi

at the beginning, information about the devices, description from the diagram:
SRV_Core - stack X670G2 (2 device) - firmware 22.3.1.4
SRV_Core1 - stack X670G2(2 device) - firmware 31.7.1.4

wsalkos_0-1703240671854.png

 

I want to do MLAG between two X670G2 stacks.

currently they are connected to each other in sharing via 4 QSPF ports where all vlans are tagged.

I checked this article about creating mlag:

https://extreme-networks.my.site.com/ExtrArticleDetail?n=000005334

and it doesn't seem complicated, I will use the existing LACP connection between stack, add an ISC vlan there and configure the relationships between peers.

Currently, the servers and my firewall cluster with gateways are connected to the SRV_Core stack. It has version 22.3.1.4 because I can't even update it because it will break.
Therefore, I want to add two more switches (SRV_Core1) that work in a stack and make MLAG between SRV_Core and SRV_Core1.

Then I want to reconnect devices from the SRV_Core_Node2 switch to SRV_Core1_Node1, so as to release devices with the name Node2. Then, if one stack is updated, only one uplink will drop.

on srv_core, for example, I have a firewall connected to ports 1:35 and 2:35, which are l2 lacp. now I would like to transfer port 2:35 to the second SRV_Core1 stack to port 1:35.

I want to do it all the time, the question is, is it possible?

I understand that I will have to add the same mlag ID to ports 1:35,2:35(SRV_Core) ,1:35,2:35(SRV_COre1), but will this process break the connection between the currently running SRV_Core and the Firewall?

i.e. what I would do.

I will run MLAG between the stack, it probably shouldn't generate a break - add that SRV_Core1 will not have a thread connected beyond lacp sharing to SRV_Core.

then on SRV_Core I will do:
enable mlag port 1:35 peer SRV_Core1 id 10 (will this generate a break if I now have 1:35 and 2:35 in L2 lacp?)
enable mlag port 2:35 peer SRV_Core1 id 10

I will then move to the second SRV_Core1 stack:
enable mlag port 1:35 peer SRV_Core id 10
enable mlag port 1:35 peer SRV_Core id 10

and then I can reconnect port 2:35 from SRV_Core to 1:35 SRV_Core1?

and the second question about connecting the server

I have ESXI servers connected to SRV_Core, the servers have two 10G cards that are connected to Node1 and Node2 of the stack but WITHOUT LACP.

For example, I have a server in SRV_Core, port 1:10 and 2:10 without lacp - some teaming works on the ESXI server.

now if I want to reconnect port 2:10 to SRV_Core1 to port 1:10, do I have to run LACP on the ports? I found information that the best practice is to run lacp on the downstream port, only then I have to reconfigure all ESXI hosts?

2 REPLIES 2

Stefan_K_
Valued Contributor

@wsalkos wrote:

I want to do it all the time, the question is, is it possible?

I understand that I will have to add the same mlag ID to ports 1:35,2:35(SRV_Core) ,1:35,2:35(SRV_COre1), but will this process break the connection between the currently running SRV_Core and the Firewall?

You can do it like this.

then on SRV_Core I will do:
enable mlag port 1:35 peer SRV_Core1 id 10 (will this generate a break if I now have 1:35 and 2:35 in L2 lacp?)
enable mlag port 2:35 peer SRV_Core1 id 10

Second command is not needed as this is a Load-Sharing member port.

I have ESXI servers connected to SRV_Core, the servers have two 10G cards that are connected to Node1 and Node2 of the stack but WITHOUT LACP.

For example, I have a server in SRV_Core, port 1:10 and 2:10 without lacp - some teaming works on the ESXI server.


Then you can just connect the ESX-Hosts to SRV_Core and SRV_Core1 without MLAG or LACP. It's not required.

Just don't forget to add all VLANs on the Links between SRV_Core and SRV_Core1

Stefan_Sleik
New Contributor II

Hi,
i fight with a similar problem the last week, so i post my Solution, and meybe it helps you (and others) to.
First read this (To understand what the Parameters will do): 
https://extreme-networks.my.site.com/ExtrArticleDetail?an=000079895
After that ther is a good howto in German: 
https://www.krakovic.de/extreme-networks-mlag-grundlagen-beschreibung-konzept/#more-12438

When the MLAG is up and Running and you like to configure a Downstream port use silimar like these:

Start orchestration mlag MLAG-ISC
enable sharing 1:10 grouping 1:10 algorithm address-based L3_L4 lacp
Enable mlag por 1:10 peer MLAG-ISC id 10
Configure vlan red add por 1:10 tagged
Stop orchestration

In Your Case ther sharing algorythm will be the point. I think it has to be the Same on the Switch as well on the VM-Ware. Is Someone ther out who can deliver details about that?

 

Best Regards
Stefan

 

 

 

 

GTM-P2G8KFN