VM randomly dropping it's NIC connection

[Vodder]

Hi,

I have a bizarre funny on one of our hosts. It's on it's own and hosts 8 VM's. However when rebooting a VM it sometimes can't talk back out to the external network and other servers cannot ping it. If you login to it the NIC is connected but it cannot ping it's gateway address. If you edit the settings of the VM and change the NIC from connected to disconnected and then back again - "in effect reseating the network cable" - the VM can then ping the it gateway and thus other servers in our network can then see it again.

The strange thing is this doesn't happen all the time and seems to effect the VM's randomly and doesn't always happen when it is rebooted. Each VM was built from the same windows 2003 template but so were other VM's on other hosts without hiccup. What was worrying is that this happened last night to one of the VM's even though it wasn't rebooted.

Any ideas? I've checked the net and can't find anything and the logs don't seem to point to anything obvious.

I do have a vimdump file stating "[Errno 25] Inappropriate ioctl for device" but doesn't say what VM this refers to

Thanks :smileygrin:

[kjb007]

If it's only happening on one of your hosts, I'd check the physical switch and make sure port security is not causing your vm to stop responding. If your vm is using one port before the reboot, and then shows up on another when it comes back up, then port security might shut it down. May not be the root cause as you are seeing an additional error, but it's one thing to rule out.

-KjB

[Vodder]

I ask them but check but I would have thought if that was the case each VM on the host would be effected as they all go out the same 2 physical NIC on the virtual machine network.

I built a new VM this morning and had the same problem. Netowrk is fine for a bit and then reboot and it can't ping its own gateway. Once the nic is disconnected then reconnected to the VM it all suddenly works. I've dumped the diagnostics but here is so much it's like looking for a needle in a haystack and not knowing what the needle looks like!

[kjb007]

Have you left the vm with the disconnected for some time after the reboot to see if it ever comes back up on its own? Is this a windows vm? Have you tried to uninstall/reinstall the tools?

-KjB

[Vodder]

Most it's been left was over a weekend and IF I remember correctly it came back on it's own accord. Yep these are Windows VM's and the tools have been reinstalled. I've "cheated" now and logged a call ?:|

[mylvisaker]

I've also experienced this odd behavior. I have not coodinated it to a particular host/port/blade - I'll look at that next time it happens. When we upgraded to 3.5U2 the vm tools were upgraded...and it seems to happen less often. But that may just be my perception. Please post the results of your support call.

Mark

[Vodder]

Will do

[Aladen]

Check how many VM's you have running, and how many ports you have configured on your virtual switch. I had a similar problem, and traced it to running out of ports on the vswitch.

No error messages were generated that I saw. Just some VM's would not get network connections on power on or vmotion.

Increased the ports per vswitch, reboot and it all worked fine.

[Vodder]

Hi, thanks for the message but unfortunately it's not that. It's a 56 port vswitch and currently only 14 VM's. Still working with VMware about solution, waiting on our Netowrk team to confirm if the physical NICs are all in the same VLAN.

Cheers.