Unreliable HKG peer with persistent UDP connections

We’re trying an experiment with using ZT/Warp+ to connect to a srcds server, which primarily uses UDP. Since the closest peer to me (MNL) is re-routed to HKG, HKG seems very unreliable as it registers a high amount of packet loss.

image

I have never encountered this much packet loss in the nearest peer, which seems to stem from my ISP’s poor interconnect in Hong Kong. Would be nice if the Warp client can catch this and ask for a better route.

Have you run a traceroute or MTR to find out which hop is dropping all the packets?

I did some tracert beforehand it didn’t seem apparent which hop was causing issues, but it seems the connection near me is stable.

I re-ran tracert again and I found out nothing abnormal so something is absolutely happening when connecting via a srcds client (any Source engine game)

This is what the netgraph inside a source engine client sees when connected via Warp+
image

The percentage of packet loss shown in your first screenshot(warp client) isn’t really high. Less than 0.1% packet loss shouldn’t cause issues. Also, a traceroute isn’t a really good way to check packet loss. You should run a MTR to the destination IP with a application called WinMTR

Heya, after running WinMTR it did give me an idea which peer was having problems. Looks like the second peer in the traceroute was having 3% loss.

image

Packet loss at intermediate hops doesn’t matter. The only hops that matters is the final hop and there is no packet loss on the final hop. Also, you don’t seem to be connecting to the nearest warp colocation server, assuming there is colocation server in your country. Which colocation server is warp connecting you to and which country are you from?

The colocation country I’m nearest is the Philippines, which has two nearest peers in my area, however it seems Cloudflare skips those peers.

However I do think despite packet loss in the second hop, it shouldn’t intermittently cause network interruption warnings in the client but it does, so there’s definitely something broken in the route.

It will be due to the routes chosen by your ISP to Cloudflare. Likely your ISP is choosing a cheaper or less congested route.

I’ll check in with the ISP tomorrow, it’s really not normal for Warp to drop packets.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.