Hello,
Long story short - when I have dvuplinks in Active/Active or Active/Passive, I get connection drops from the hosts connected to the switch. When only one dvuplink is used, all is right with the world.
A couple of months ago, I built out a new three-node cluster (ESXi 5.1 U1 b1312873). I built a 5.1 vDS and configured it using default load balancing (as well as attempting with Route based on physical load & explicit failover order). For Network Failover Detection, I've chosen Beacon Probing as per the direction of the network engineers.
Each host has two network ports (for a total of six nic ports used) for:
Management/Server LAN (0 and 1)
NFS (2 and 3)
vMotion (4 and 5)
Each pair of host NICs is plugged into a separate 2K switch...for physical switch FT. NICs 0,2,4 go to one 2K, NICs 1,3,5 go to the other 2K.
Our physical networking setup is two Cisco 5Ks with two fabric extender 2Ks...one off of each 5K. In our current setup we have ESXi hosts dual-homed on the nexus 2k’s. When the dv uplinks are brought up in active/active or active/passive on the applicable vDS port group, we receive periodic ping drops. In order to avoid this problem we have limited the connection to using only one of the dv uplinks.
Here's the SW software version for the Cisco switches:
Software
BIOS: version 3.5.0
loader: version N/A
kickstart: version 5.1(3)N1(1)
system: version 5.1(3)N1(1)
power-seq: Module 1: version v1.0
Module 3: version v2.0
uC: version v1.2.0.1
SFP uC: Module 1: v1.0.0.0
BIOS compile time: 02/03/2011
kickstart image file is: bootflash:///n5000-uk9-kickstart.5.1.3.N1.1.bin
kickstart compile time: 12/6/2011 22:00:00 [12/07/2011 01:30:01]
system image file is: bootflash:///n5000-uk9.5.1.3.N1.1.bin
system compile time: 12/6/2011 22:00:00 [12/07/2011 03:09:44]
If anyone has any thoughts, ideas, feelings....I'm open to suggestion. Thanks so much in advance!
