ESXi

 View Only
  • 1.  Cluster question

    Posted Apr 12, 2010 03:07 PM

    This weekend a few VMs on a clusered ESX Dell 2950 lost networking for some reason. The console and VMotion were fine. The 2950 has 4 NICs. I have two for the console and VMotion. And I have two for the VMs. The VM networking is teamed into a Cisco 3750 gig switch. Looking at the host in question I'm missing one of the NICs for the VM network. Looking into that. Taking the host down and looking at the BIOS to see if the NICs are all present. I have activity lights on all NICs. Two of the NIC are on board and the other two are an add-in card. I used one NIC onboard for VMs and Another NIC from the add-in card. This way the VMs don't runn off the same networking hardware. It odd the one NIC showing for the VMs didn't pick up the load. Might be an issues on the Cisco 3750 but the other hosts are fine.

    Leading me to understanding clusters. I was able to vmotion the VMs to a working host. The last one said it moved but it really didn't. tried again and it said it moved. But under summary it showed on the bad host. I had to remove it from inventory and add it on the good host to get it back up. Shouldn't the cluster sense and issue and try to move the VMs to a known good host instead of me needing to do it manually?

    Craig



  • 2.  RE: Cluster question

    Posted Apr 12, 2010 03:29 PM

    I did find out the Dell 2950 has a missing on board NIC. Getting that taken care.



  • 3.  RE: Cluster question

    Posted Apr 12, 2010 03:48 PM

    I think you have found your problem - in terms of HA it will only kick in if the host is down or if your monitoring per VM the ESX host has lost the VM heartbeat not for networking issues -

    If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful



  • 4.  RE: Cluster question

    Posted Apr 12, 2010 05:42 PM

    What I have to find out is why the other NIC didn't take over the load. It appears it is up and running. Going to work with it until Dell takes care of the failed NIC.

    Craig



  • 5.  RE: Cluster question

    Posted Apr 12, 2010 06:10 PM

    What failure sis you have selected? If it was Link State and NIC still had an active link light but there was a problem with switch configuration than the the traffic would not have failed over - Link State + Beaconing would have detected the failur e-

    If you find this or any other answer useful please consider awarding points by marking the answer correct or helpful