Hyper-V high latency on gateways reported
-
Dear all,
i have been using the nightlies of the Pfsense for Hyper-V (using 2012 R2 now) since the 2.0 release and always had this problem: very high ping of the gateway either in the GW itself or a different monitor IP. Before the 2.2 release it was around the 40k now it is around 5k. See pic attached. With one WAN it would be easy as i would mark the line as always up but now with several WAN's, sometimes the WAN's would be marked as down.
This actually defeats the purpose of having GW groups and rules.
Is there a fix for this or at least an idea of what to search for etc?
![Gateway latency.png](/public/imported_attachments/1/Gateway latency.png)
![Gateway latency.png_thumb](/public/imported_attachments/1/Gateway latency.png_thumb) -
We have the same problem. yesterday we moved from a dedicated computer to a hyper-v machine and the gateway's latency jumped high.
Did you find out what to do?
I've found this thread: https://forum.pfsense.org/index.php?topic=76481.0. but haven't checked that out (waiting for the night).or maybe it related to the network card and we need to move to intel network card.
Thanks,
Nadav. -
No need for Intel cards. I had about 5-6 types of Intel NIC's included 10GBps X540 ones + several other brands. Makes no difference is it's Realtek, Boardcom etc.
I have asked a fellow admin to look and he made only one modification to the virtual switch. Marked with red in screenshot I did not want a shared network with the OS for the GW. Now the latency is within normal values but this is quite new i would give it a few more days to confirm it.
![latest mods.png](/public/imported_attachments/1/latest mods.png)
![latest mods.png_thumb](/public/imported_attachments/1/latest mods.png_thumb) -
The timing in Hyper-V FreeBSD VMs can be really bad. As soon as that gets fixed upstream, we'll pull it in. It's an issue Microsoft is aware of and is on their road map to fix. Generally safe to ignore as it tends to not impact anything other than what gateway latency is reported as.
I don't think that checkbox had any relation. It's something that comes and goes, probably just a coincidence that it's gone away (for now at least) after doing that. But definitely interested in hearing your experiences, follow up once it's been running for a while.
-
maybe it related to window's nic teaming.
I found that without nic teaming it was better. -
@cmb:
Generally safe to ignore as it tends to not impact anything other than what gateway latency is reported as.
so apinger doesn't failover to another wan in those situations ? i figured those number were fetched from apinger.
-
apinger is one giant piece of beep
-
found a solution. I disabled the Time Synchronization in Hyper-V's guest settings and restarted the gateway and it works fine.
-
found a solution. I disabled the Time Synchronization in Hyper-V's guest settings and restarted the gateway and it works fine.
thanks man, you saved me a lot of time :D
-
Yes, any gateway latency reporting issues that still exist in latest versions (and most in older versions) with Hyper-V are because you have both NTP enabled inside the VM, and Hyper-V's time sync enabled. That does bad things to the system clock, which makes gateway monitoring inaccurate. Disable NTP or Hyper-V's time sync.