Hyper-V Server runs like treacle, no RDP or IPV6 settings

[Warwick_Tech]

Hi All,

 

A Hyper-V server has suddenly decided to run like treacle, and refuse RDP connections

 

Though wierdly the background processes run fine, as well as the host so It just seems to be a GUI issue (?)

 

I've done my digging and the only diffrences I can find with this machine is some missing IPV6 settings compared to others:

DHCPv6 IAID

DHCPv6 Client DUID

 

these are missing on the server

 

Memory is set at 35000 MB, plenty of HDD space, same network adaptor as the others -

 

Processors: 8

numa nodes: 1

Sockets: 1

Hardware threads per core: 2

 

Numa topology -

Max processors 16

max memory 46948 MB

max numa sockets 1

hardware threads per core 0

 

NUMA spanning is ON

 

Do I need to follow Jeremy Clarksons advice and just give it more POWER

[Michael]

Have you looked at iLO or iDRAC, depending if it's a HP or Dell server? What's the uptime? A reboot may clear the issue, as it sounds software related.

[Warwick_Tech]

Have you looked at iLO or iDRAC, depending if it's a HP or Dell server? What's the uptime? A reboot may clear the issue, as it sounds software related.

 

It's a bespoke build on a S2600GZ Mobo - I don't think ILO is plugged in, not heard of IDRAC

 

Also rebooted less than 24hrs ago and ran an sfc /scannow and it's still at 56%!

[Michael]

iLO and iDRAC are the same thing, just different manufacturers. I personally favour HP's iLO between the two.

 

It does sound something software related if you're running sfc /scannow, presumably on the Hyper-V host?

[Warwick_Tech]

Well, the host runs fine, RDP and slick - It also hosts two other machines that are running fine...

 

I'm looking at Numa modes but tbh I'm not sure if changing these will do anything (?)

[Michael]

Well, the host runs fine, RDP and slick - It also hosts two other machines that are running fine...

 

I'm looking at Numa modes but tbh I'm not sure if changing these will do anything (?)

 

So is this part of a larger Hyper-V cluster? It sounds that way.

[Warwick_Tech]

Yes, but just the one VH and 3 VM's, physical resources are roughly a quarter for each machine (1/4 RAM 1/4 Processors)

[slugshead]

The beauty of a cluster is that if one runs like crap (as op states) you can just bring it offline, fix/rebuild quickly and easily or even add additional nodes.

 

live migrate your VMs to another host and rebuild the dodgy host. it's probably quicker to do that than try repairing etc. Unless that's not an option?