11-29-2013 02:21 AM
the customer has problems with VC and MS NLB. Did anybody faced similar problem?
- Bladesystem c7000, BL460 g8 servers, VC FlexFabric 10Gb/24-port (4.10 f/w) connected to Cisco 6500 (VSS), VMware vSphere 5.1 cluster inside chassis.
- Active-Active VC Config (Dual SUS), LAGs of 2 links to Cisco6500 (LACP), now only 1 VC active for diagnostics
- 2 x Windows 2008R2 VMs in NLB cluster (multicast mode) in VLAN X (inside chassis)
- Terminal server VMs in VLAN Y (inside chassis)
Traffic VLAN X - VLAN Y is routed by Cisco 6500, recommended settings for NLB (static arp etc) are done
When 2 links in LAG from VC active traffic does not go from Terminal server to NLB cluster vip (session drops etc).
For clients outside c7000 chassis everything is ok.
When only 1 link is active from VC everything is ok (does not depend on which link is active).
Only multicast traffic is affected.
06-04-2014 01:12 AM
So sorry to dig up such an old post, but I have a scenario which is very similar.
The exceptions to this are, customer is using Hyper-V and traffic in the whole VLAN was experiencing issues until the servers configured in the NLB were shutdown.
Can anyone provide any guidance on where we go with this?