Ping issue when "Destination host unreachable"

Meserias

hello all,

When I connect to server, everything works as expected and I can reach any machine via VPN from client to server's LAN....
Performance it's where it should be and if I ping or access any server side machine via VPN everything works perfectly.

But, when I ping via VPN a machine from server's side LAN which is not started (been shutdown or power off because it was needed) all VPN channel connections and performance are slowed down, and also other new VPN connections are heavily affected by the ping which is NOT reaching that powered off station.

The ping that I initiate, it's normal with no special options:

C:\ping 192.168.62.3
Reply from 192.168.62.26: Destination host unreachable.
Reply from 192.168.62.26: Destination host unreachable.
Reply from 192.168.62.26: Destination host unreachable.
(of course I understand it's should not show anything else - because station it's OFF - but why VPN channel performance drops ?)

Now I have the following question:
what options should I need to add to server or client in order to not been affected by this.
to me it's looking like a buffer it's filling up or something

TinCanTech

Moved to "Server administration"

Openvpn simply transports the ping packet from the client to the server, where it is handed off to the server network stack to be processed.

It sounds to me, more like some other issue taking place on the server, which happens to look like what you describe.

Without configs and debugging logs, there is no way to diagnose this, except by a wild stab in the dark.

Please see: https://forums.openvpn.net/viewtopic.php?t=22603

Meserias

Can someone please test this ? This is actually dead simple...
from a Windows client just ping a non-alocated (not in use) IP from the server's LAN area.
(let it ping a while and test and examine how your OpenVPN performs while PING it's failing as this is expected)

NOTE: Do not forget in my above setup everything it's working, and I have strong, stable, good connection via OpenVPN.

TinCanTech

Meserias wrote Can someone please test this ?

No. The problem is not what you think it is.

If you do not document your VPN installation then we cannot help. See the link above.

Meserias

maybe someone else, not you, especially. Thanks
I understand that you refuse to believe that anything else in my network it's working as expected
(including pinging a station which is up and running !)

TinCanTech

You have not documented your setup, we don't even know your server operating system.

Ciao..

Meserias

Server it's Windows, client it's Windows too... I have under 1ms trough OpenVPN to a location about 25km away 🙂)))
I know that it's very good for this distance in a crowed city.
I think I have found an explication OpenVPN it's single threaded as PING filling the buffer connection gets more and more slow......
and a workaround to this is to just increase buffers for send and receive.
I'm not too happy with this, but this is what it OpenVPN, I hope OpenVPN gets multithreaded process in the future.

ordex

Hi,

Meserias wrote I think I have found an explication OpenVPN it's single threaded as PING filling the buffer connection gets more and more slow......
and a workaround to this is to just increase buffers for send and receive.

what buffer are you currently referring to? can you point it out?

Also, can you explain how you worked around this issue?

Thanks!

Pippin

buffers for send and receive

--rcvbuf and --sndbuf
https://community.openvpn.net/openvpn/ticket/461

Meserias

you need to test for yourself and play with settings.
Lucy me, that this is not business critical for me.
Multithreading OVPN should be implemented.

Pippin

ordex wrote what buffer are you currently referring to? can you point it out?

Also, can you explain how you worked around this issue?

Thanks!

Can you answer the questions?

ordex

Pippin wrote
buffers for send and receive
--rcvbuf and --sndbuf
https://community.openvpn.net/openvpn/ticket/461

Thanks for this.

Did you find a way to workaround this in the meantime?
I am not sure about multithreading, because this is not something that happens on other platforms (i.e. on linux).
The host unreachable concept is something that is generated outside of OpenVPN, when the packet has been passed to the system networking stack. Therefore it is kinda weird (but not impossible) that is OpenVPN to slow everything down.

Maybe it's related to tap-windows6. Is this that you are using?