SRX Services Gateway
Reply
Regular Visitor
marks
Posts: 4
Registered: ‎03-14-2012
0

SRX 240 Cluster SWFAB Issues

I attempted using the new swfab feature for ethernet-switching in SRX240 clusters, but had a few issues, one of which was a showstopper.

 

First, loops were created across the cluster until I implemented spanning tree on the cluster (very strange since the ethernet switching process should only be running on one node at a time).  

 

Second, and the reason I can't use this in production, is that when I tested failover by pulling the power on the primary node, the second device failed over but ethernet switching never worked until the original node was powered back up and the swfab interface showed up.  

 

This is running JUNOS 12.1r2.9 on the SRX's.

 

I did notice the note that

 

"Behavior of the nodes may be unpredictable, if the Ethernet switching-related features are configured before configuring the swfab interface on both the nodes."  

 

However, I made certain to enable the swfab, etc options, commit, reboot, then went on to configure additional items, like family ethernet-switching, lacp, etc.

 

My demo kit consisted of 1 EX2200 and 2 SRX240's in HA.  I've included a diagram for reference.  The looping starts as soon as I plug in the "laptop host" and the broadcast can be seen on all three lacp interfaces on the srx's as well as on the ex2200.  

Visitor
Scapal
Posts: 6
Registered: ‎12-22-2011
0

Re: SRX 240 Cluster SWFAB Issues

I toyed with that as well in january, came out with the same results as you did. 

I used an early 11.4 release back then.

 

Have you tried this again with a recent release, this would save me the trouble of testing this again, feeling like a monkey in a space rocket.

Contributor
Fabian Prou
Posts: 10
Registered: ‎07-19-2011
0

Re: SRX 240 Cluster SWFAB Issues

Hi Mark, You are talking about LACP interfaces? AFAIK, LAG/LACP is not supported in Cluster mode when a single LAG group has links connected to both SRX. Here the note SRX 12.1 : NOTE: The redundant Ethernet interface LAG child links from each node in the chassis cluster must be connected to a different LAG at the peer devices (That's is your EX2200). If a single peer switch is used to terminate the redundant Ethernet interface LAG, two separate LAGs must be used in the switch. Regards, Fabian.
Copyright© 1999-2013 Juniper Networks, Inc. All rights reserved.