VMware Cloud Community
Arlathen
Contributor
Contributor

Mysterious Guest Network Disconnects

Hi All,

My first post here, of course the mother of necessity means that I'm posting a problem I have instead of trying to be helpful to others:)

We're running a Vcentre / ESXi estate on a HP BladeCentre / HP 3Par Storage with primarily Windows Server 2008 Guest VMs. Very recently have noticed that a small number of Guest VMs may disconnect from the network, losing network access.

Our VMware estate has under-gone significant change over the last three months - we've upgraded Vcentre from 5.5 to 6.5 U1, and then upgraded our Hosts to 6.5 U1 as well. We've also recently installed and licensed NSX, as we're looking to use the firewalling aspects along with Vrealize Network Insight to add some network separation to our security improvements.

To fix the issue, we simply disconnect the network card in VCentre and then re-connect it, or occasionally reboot the server, and hurray the server is back on the network. The issue has started to escalate however and we've now seen a total of 7 or 8 VMs display this behaviour in business hours.

Some thing's we've looked into:

  • Its affected Guests with both E1000 network adaptors and VMXNET3 adaptors.
  • It's affected Guests with different VM Hardware versions, with versions as high as VM hardware v11
  • It's affected Guests with different versions of VMTools,
    • In all three of these cases, we're trying to get all 200+ VMs moved upto VMXNET3 and the latest HW / Tools versions.
  • We do have a separate management and production clusters, with dedicated Blades serving each
  • We use Distributed switching through out our deployment
  • We don't have Jumbo frames enabled

I've tried reviewing both the VMware.log for the Guest itself and also the various logs for the Host via VRealize Log Insight, as we have that deployed to support our NSX installation going forward. I did try to manually look through the text logs after one particular incident but they only went back 30 minutes and missed the incident period Smiley Happy

Can anyone suggest some starting other points for troubleshooting the issue?

Thanks for the feedback!

0 Kudos
1 Reply
mhampto
VMware Employee
VMware Employee

This can happen due to multiple reasons. Please see VMware Knowledge Base  for further troubleshooting and finding a root cause.

0 Kudos