Extreme networking issues, virtually unplayable in maps, problems to PoE server route

Hello everyone,

I currently play on Texas, or California servers but I am located in Denver Colorado. The game is virtually unplayable in either lockstep (insane stutter or multi-second freezes as game waits for server info) or insane rubberbanding with predictive networking.

I've ran a winMTR for path of exile, ran a typical map, and found one of the hops is completely unresponsive, and the 6th hop is showing 18% packet loss. That IP is for a datacenter in Denver, but this seems like an issues for people worldwide.

Whats going on here?



|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| *** - 0 | 1974 | 1974 | 0 | 0 | 17 | 0 |
| ***-dsl-gw17.hlrn.qwest.net - 1 | 1963 | 1960 | 0 | 5 | 108 | 4 |
| ***-124-129.hlrn.qwest.net - 1 | 1971 | 1970 | 0 | 5 | 50 | 4 |
| No response from host - 100 | 397 | 0 | 0 | 0 | 0 | 0 |
| ae2.3602.edge8.Denver1.net.lumen.tech - 1 | 1951 | 1945 | 0 | 7 | 70 | 17 |
| 4.71.42.142 - 18 | 1165 | 960 | 0 | 9 | 111 | 5 |
| 172.68.32.12 - 1 | 1952 | 1946 | 4 | 7 | 77 | 5 |
| 172.65.166.147 - 1 | 1956 | 1951 | 0 | 5 | 45 | 5 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider

I am having the same issue as you describe, seemingly regardless of what gateway I use... I made a pastebin thingy for a bad instance I had: https://pastebin.com/UEhJ59Ym

I play from Denmark, and changing gateway to another EU-based one, has almost never worked in the 7-8 years I've been playing. I recently tried changing to the Texas gateway, and that seems to sometimes resolve my issue (though having 150 ping), but not in any consistent manner.

EU servers have been bad for years, but these past two weeks have been rough. I have been almost completely unable to play the game today.

I play on Texas servers typically, but I am routing from Denver, which isn't far by "US standards". I am going to call my ISP tomorrow but I got a feeling they are just going to run a speedtest and call it a day.

I ran PingPlotter and there are multiple packet loss issues at multiple locations for data center in my networking path to PoE servers.

My computer runs the game fine, I am at 90fps in maps while the game functionally operates at 10fps with extreme stutters or rubberbanding depenind on which networking mode I use.

I'm also in Denver and having a very similar issue. It gets worse with mob density. Trying to play deli/breach is currently a very weird slideshow and then everything catches up at the end of the map.

This is my exact experience. Absolute slideshow but the game running at 80+ fps clientside. It seems like it catches up at the end but not in a good way because them everything is running around cracked.

"

Endorsmint#1120 wrote:

| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| *** - 0 | 1974 | 1974 | 0 | 0 | 17 | 0 |
| ***-dsl-gw17.hlrn.qwest.net - 1 | 1963 | 1960 | 0 | 5 | 108 | 4 |
| ***-124-129.hlrn.qwest.net - 1 | 1971 | 1970 | 0 | 5 | 50 | 4 |

The part of the WinMTR we actually got to see shows packet loss beginning immediately after the router in your home when your traffic first reaches your ISP. There are no other signs of an issue.

It is possible performing a power cycle could be helpful.

https://www.poecommunity.help/miscellaneous/other/perform-a-power-cycle

Here is another WinMTR, it's showing 37% packet loss from IP 4.71.42.142, which is a data center in Denver. I power cycled and left the modem and router off for an hour yesterday. Still same issues today.



|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| **** - 0 | 359 | 359 | 0 | 0 | 17 | 0 |
| ***.hlrn.qwest.net - 1 | 356 | 355 | 0 | 6 | 61 | 3 |
| ***.hlrn.qwest.net - 0 | 359 | 359 | 3 | 5 | 43 | 4 |
| No response from host - 100 | 73 | 0 | 0 | 0 | 0 | 0 |
| ae1.3502.edge8.Denver1.net.lumen.tech - 5 | 312 | 299 | 0 | 7 | 28 | 11 |
| 4.71.42.142 - 37 | 147 | 93 | 0 | 10 | 125 | 5 |
| 172.68.32.12 - 1 | 356 | 355 | 0 | 8 | 70 | 6 |
| 172.65.166.147 - 1 | 356 | 355 | 0 | 5 | 17 | 5 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider

"

Endorsmint#1120 wrote:

it's showing 37% packet loss from IP 4.71.42.142, which is a data center in Denver

It isn't a data center, and the 37% packet loss reported on that hop isn't relevant.

"

Endorsmint#1120 wrote:

| 172.65.166.147 - 1 | 356 | 355 | 0 | 5 | 17 | 5 |

The output from the test shows 1% packet loss to the destination.

The WinMTR log which you insist on redacting for reasons that aren't entirely clear first reports 1% of packet loss on hop #2. While the following hop doesn't report any packet loss, it is common to see some degree of variance and a decrease of 1% certainly isn't significant in that context.

It is possible that running the test again for longer would clarify whether hop #2 really is just coincidentally dropping some packets but is not connected to the packet loss seen later - which seems like quite a coincidence, but hasn't technically been ruled out thus far - or if hop #3 not dropping any of the 359 echo requests sent to it during the test was simply a fluke (which would be my guess).

As an aside - are you running this against sjc.login.pathofexile.com? The target appears to be a Cloudflare device rather than one of GGG's actual instance servers. If you are running the test against the wrong company's infrastructure in a different data center the output from the test won't necessarily be reflective of your in-game experience.

"

Sarno#0493 wrote:

"

Endorsmint#1120 wrote:

it's showing 37% packet loss from IP 4.71.42.142, which is a data center in Denver

It isn't a data center, and the 37% packet loss reported on that hop isn't relevant.

"

Endorsmint#1120 wrote:

| 172.65.166.147 - 1 | 356 | 355 | 0 | 5 | 17 | 5 |

The output from the test shows 1% packet loss to the destination.

The WinMTR log which you insist on redacting for reasons that aren't entirely clear

How is it not relevant? If it's not a data center, what is it? I guess it doesn't matter.

Please explain to me why you need the full addresses and I will provide them. Who are you again? GGG doesn't provide tech support so are you just hanging around their forums? If you are trying to help people, perhaps being passive aggressive on the internet isn't going to be particularly effective.

Yes I am running the WinMTR against sjc.login.pathofexile.com.

As others in the Denver, Colorado area have noted, this doesn't seem to be a singular issue. I also notice latency issues outside of the game (streaming, youtube for example).

"

Endorsmint#1120 wrote:

Here is another WinMTR, it's showing 37% packet loss from IP 4.71.42.142, which is a data center in Denver. I power cycled and left the modem and router off for an hour yesterday. Still same issues today.



|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| **** - 0 | 359 | 359 | 0 | 0 | 17 | 0 |
| ***.hlrn.qwest.net - 1 | 356 | 355 | 0 | 6 | 61 | 3 |
| ***.hlrn.qwest.net - 0 | 359 | 359 | 3 | 5 | 43 | 4 |
| No response from host - 100 | 73 | 0 | 0 | 0 | 0 | 0 |
| ae1.3502.edge8.Denver1.net.lumen.tech - 5 | 312 | 299 | 0 | 7 | 28 | 11 |
| 4.71.42.142 - 37 | 147 | 93 | 0 | 10 | 125 | 5 |
| 172.68.32.12 - 1 | 356 | 355 | 0 | 8 | 70 | 6 |
| 172.65.166.147 - 1 | 356 | 355 | 0 | 5 | 17 | 5 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider

"

Google LLM wrote:

The 37% packet loss at node 6 (4.71.42.142) is a false alarm (ICMP rate limiting). It doesn't affect your final connection because the final hop shows only 1% loss.
The Real Issue:
1% Packet Loss: This starts at hop 2 (hlrn.qwest.net) and persists to the end.
Conclusion: The problem is in your ISP’s network or their immediate uplink, not the "37%" node.
Impact: Minor stutters in gaming or VoIP; unnoticeable for browsing.

I had to kick LLM twice to get her to start giving the correct answers.

"

Endorsmint#1120 wrote:

Yes I am running the WinMTR against sjc.login.pathofexile.com.

"

Google LLM wrote:

Tracing to a Login Server instead of an Instance Server is a bad idea for three main reasons:
Physical Location: The Login server is often in a different data center (e.g., a central hub) than the Game instances, which are spread across multiple regions to reduce latency.
Network Routes: Your traffic takes a completely different path (different nodes and ISPs) to reach a game instance. A "clean" path to the Login server doesn't guarantee a stable path to the actual game world.

Bottom line: Checking the Login server only tells you if you can log in, not if you can play without lag.