New implementation of vSphere 5.1 (U1) , ESXi 5.1 (U1) running on HP c-Class Blades. The blade environment consists of C7000 chassis, 4 x Flex10D VC Modules and BL460c G7 blades. All hardware firmware versions and esxi driver versions are as per the VMware/HP April 2013 version of the Recipe Book (See attached).
To provide some of our networking functionality (NFS, vMotion, FT Logging and VM Networks) we are making use of the Cisco Nexus 1000v switches which have uplinks via 2 virtual connect modules (in bays 5 & 6)
During testing of the environment we have run into issues when testing failure of Virtual Connect modules that connect back to the 1000V. The issue is that we are getting no alerts flagged on the hosts when loosing a virtual connect module. The 1000v vsm's show that the links are down, the host networking view in the vCenter GUI show that correct vmnic's are down and SNMP traps are being generated correctly. We are just not getting an alert flagged up on the host in vCenter.
We also have some of network functionality (Host management & iSCSI) that does not pass via the 1000v's and use standard vSwitches. These networks pass through the remaining 2 virtual connect modules in bays 1 & 2. For this traffic we are almost seeing the reverse with host alerts in vCenter.
When one of the VC modules is powered down all hosts generate Network connectivity Lost and NIC redundancy alerts as expected and all traffic fails over to the remaining path as expected. The issue comes when the 'failed' VC module is brough back online. In this scenario only some of the host alerts are cleared, even after extensive checking to confirm that all network paths are back online.
No changes have been made to the default alert definitions that are configured in a standrd vCenter installation.
I have just logged a ticket with VMware but thought I would put a post out on the furms to see if anyone else is seeing similar issues.
Regards,
odee30