Randomly high ping times across 2 guests on Hyper-V virtual switch?

[rpeters83]

I have a small personal Hyper-V server with 12 guests all sharing one virtual switch. I was doing some tests to see why some pings were randomly high and I wrote a script to log/email high ping times. It runs on one of the guests and pings IPs, such as the router and other devices on the LAN.

Out of curiosity, I had this guest ping another guest on the same server. In theory, I thought that this would go across the virtual switch without even leaving the physical server, but I've noticed randomly high ping times.

So, Guest 1 is pinging Guest 2 on the same Hyper-V server, same virtual switch, and some ping times are randomly high.

1. Between these two, would a packet even venture out into the physical network, or stay entirely on the virtual switch?
2. What would explain why some pings across the same virtual network are high?
3. Could this spike be due to some sort of software latency?

I don't really have a problem, but am more curious as to why this may be. It is usually 1 ping out of about 100 that is randomly spiked. TIA.

 

[boss6021]

Any chance VM queuing is enabled? What hardware are you running, and more specifically what NICs are you running?

 

[rpeters83]

Any chance VM queuing is enabled? What hardware are you running, and more specifically what NICs are you running?

I'm using whatever the defaults are. Stock Hyper-V install and threw on guests. I'm using a built-in Intel NIC. This is desktop hardware, as this is all for personal use.

 

[boss6021]

I'm using whatever the defaults are. Stock Hyper-V install and threw on guests. I'm using a built-in Intel NIC. This is desktop hardware, as this is all for personal use.

Ok, I would recommend you have the latest Intel drivers on the host OS, I would verify that VM queuing isn't on. Those are just the initial checks. After that, specific hardware will be needed to try and determine what is going on.

 

[rpeters83]

Ok, I would recommend you have the latest Intel drivers on the host OS, I would verify that VM queuing isn't on. Those are just the initial checks. After that, specific hardware will be needed to try and determine what is going on.

It is turned on for each guest it seems. Is this the setting in question below? I thought the issues with having this enabled was for broadcom drivers. Can this be for any kind of NIC?

 

[boss6021]

That's the one! Here is a good explanation of the issue, and a possible solution for you. To answer your question, Intel NICs are prone to this issue as well.

https://www.petri.com/hyper-v-network-issues-1-gbe-nics

I'm not saying that this is your issue, but I have had it solve numerous "gremlins".

 

[rpeters83]

That's the one! Here is a good explanation of the issue, and a possible solution for you. To answer your question, Intel NICs are prone to this issue as well.

https://www.petri.com/hyper-v-network-issues-1-gbe-nics

I'm not saying that this is your issue, but I have had it solve numerous "gremlins".

Thanks. If the preferred way to turn it off via powershell or individually per guest (using the setting above)?

 

[boss6021]

Thanks. If the preferred way to turn it off via powershell or individually per guest (using the setting above)?

It's completely up to you. I prefer Powershell personally.

 

[Grimlaking]

I have a small personal Hyper-V server with 12 guests all sharing one virtual switch. I was doing some tests to see why some pings were randomly high and I wrote a script to log/email high ping times. It runs on one of the guests and pings IPs, such as the router and other devices on the LAN.

Out of curiosity, I had this guest ping another guest on the same server. In theory, I thought that this would go across the virtual switch without even leaving the physical server, but I've noticed randomly high ping times.

So, Guest 1 is pinging Guest 2 on the same Hyper-V server, same virtual switch, and some ping times are randomly high.

1. Between these two, would a packet even venture out into the physical network, or stay entirely on the virtual switch?
2. What would explain why some pings across the same virtual network are high?
3. Could this spike be due to some sort of software latency?

I don't really have a problem, but am more curious as to why this may be. It is usually 1 ping out of about 100 that is randomly spiked. TIA.

Did you run a trace? Are they on the same vlan? If you ran a simple tracert is it going to the gateway before hitting the target? Is your gateway a separate virtual appliance?

If you want to make absolutely sure your virtual environment is cordoned off then set up a gateway for your guests that is a guest appliance. Then let it connect and direct traffic out to the router as needed. If your router is your gateway and you are usng vlans then your traffic will leave your box.