Hello,
I've been having a re-occurring issue where a VM reboots (Planned) and occasionally gets a 169 link local address. Its not all the time and random amongst which VM does it and on what Hyper-V cluster. The quick fix is to just disable/enable the NIC. The addresses are statically provisioned through VMM.
Setup A:
The Hyper-V hosts used to plug into in a cisco 4510 chassis (Single MAC Table). There was never an issue then.
Setup B:
We went to new core switches (IBM 8124) 10 gig switches (Multiple independent MAC Tables). These switches were vastly underpowered and chalked the 169 issue to that.
Current Setup:
We have replaced the IBM switches with Brocade VCS switches in a logical chassis configuration (one shared MAC address table but it needs to sync amongst the switches). The issue is not as rampant but still there.
The Vswitches are configured to use teaming, switch independent with both NICsactive/active
Now that we have brocade switches that support LACP fast timeout we will be converting the NICS to LACP links. I'm wondering is the problem is related to the MAC address being tied to a port and the other switch is slow to see the change causing the other switch to see that address in use? By making it an LACP link the MAC is tied to the Logical port and it wont actually ever switch ports on the Host (if that makes sense?).
Has anyone seen this before?