PSA: The servers are fine, it is your ISP's routing that is not

If the issue was exclusively on GGG's end, then everyone would have issues. But that's clearly not the case (my own experience with an EU server has been rock solid for the weeks I've been playing).

As long as none of the complainers have provided proof - that OP already asked for, but is yet to be shown - that the issue is (more) on GGG's side than on theirs, it has to be assumed that the issue lies more with the complainers.

"

BlueTemplar85 wrote:

If the issue was exclusively on GGG's end, then everyone would have issues. But that's clearly not the case (my own experience with an EU server has been rock solid for the weeks I've been playing).

You clearly understand how this all works. Hopefully you don't work anywhere network infrastructure or computers, because that would be sad for your customers.

"

BlueTemplar85 wrote:

As long as none of the complainers have provided proof - that OP already asked for, but is yet to be shown - that the issue is (more) on GGG's side than on theirs, it has to be assumed that the issue lies more with the complainers.

there are dozens winMTR logs showing those shitty instance server routings. Nothing to do with the ISP of the gamer, just with general routing in europe. GGG cloud blacklist those known troublesome routes and reroute them or move their servers from afflicted farms etc. But that would cost money so.... no.

So no there is no "has to be assumed" here. The picture is pretty clear, there are hundreds of logs online, you could have done your research, but you didn't, how cute.

If "bad routing" can cause instance "crashes" where you lose your portals, then OK; the problem might be my ISP for all I know. I just don't see why/how that's a routing issue. And everytime I get these sick-ass latency spikes, spiking into eternity, it ends with the map instance disappearing.

But I'm no computer wiz...

"

BlueTemplar85 wrote:

If the issue was exclusively on GGG's end, then everyone would have issues. But that's clearly not the case (my own experience with an EU server has been rock solid for the weeks I've been playing).

Unless you got some inside information or some other kind of information that I dont have: Not necessarily so. They may be (for example) distributing players according to some only-known-to-them-or-not-even-them scheme onto their servers for example. If they have limitations (obviously they have) then for performance optimization and such it could make sense to do that. If I would be responsible for that kind of tech the question would be expected cost of implementing a not-fully-random system vs expected benefit. There are a lot of variables available to use like crashes, load, bandwith, packet losses, instance size, expected instance duration, ...

"

BlueTemplar85 wrote:

As long as none of the complainers have provided proof - that OP already asked for, but is yet to be shown - that the issue is (more) on GGG's side than on theirs, it has to be assumed that the issue lies more with the complainers.

If not all is solely on GGGs side then there are multiple issues. Disappearing portals and rollbacks are GGG side obviously - neither players nor connection with players have control or access to that.

"

Phrazz wrote:

If "bad routing" can cause instance "crashes" where you lose your portals, then OK;

"Bad routing" with player triggering instance crashes is only possible if their software is bad. So whether this triggers it or not its still on GGG.
What could happen for example is that due to packet losses they end up in an ever-escalating resend-loop. I have seen evidence suggesting that they have a bad lockstep/sync algorithm that sends superfluous information to the player. But without inside knowledge this is speculation.

"

Phrazz wrote:

And everytime I get these sick-ass latency spikes, spiking into eternity, it ends with the map instance disappearing.

A single unending latency spike with instance-gone at the end is likely the symptom of a server failure.

Routing is a real thing but it only affects very few amount of people.

What is happening here is when you join through vpn you are simply placed to a different data center which is not on fire. Let's say you are joined through Frankfurt gateway but that does not mean there is only frankfurt datacenter in Germany. They have datacenters in multiple cities in germany. You can check the IP you are connecting to and look it up a site.

Even between maps you are sent to different ones. Hell even your hideout is in different data center than the map is you are joining. Sometimes the hideout is not lagging but the map is. It is simply because the datacenters are in constant flux. If it is full you are sent to the next one. Many many times I'm placed on a NA server but I'm on EU. Gateway still says I'm let's say on frankfurt but the IP I'm joined is located in USA.


Blaming it on player's ISP routing is pointless. Most of the ISP has great routing in their own country then what is happening outside of it is out of their reach and cannot be fixed by the provider. You can't expect everyone to buy vpn to potentially mitigate the issue.

Real talk here. Even if it was routing problem you would have 40 ms instead of 20 ms (since the signal has to travel probably double the distance or more) but in this game it lags like hell which is obviously not routing issue. THe hosts they are renting runs dozens of services other games not just poe. If it was a big issue region wide don't you think it was not adressed by now? There are far more important and bigger things ran on the same hosts at the same time than poe. Plus every time the game lags just try other online games and you will see they are working fine. Suddenly everywhere else your routing is fixed but only for poe it is bad. I understand renting servers aren't cheap nowadays and poe is famous of having small budget for servers but stop blamig it on things that is not causing the real issue.

They could add more gateways that would spread the load that could potentially fix the issue. Add a new gateway to an eastern EU country maybe to romania or hungary or something so quarter of your eu playerbase won't be joined to the west EU places thus lifting a bunch of load from london/milan/amsterdam and frankfurt of course. Basically if you are from eastern europe or balkans your best option is frankfurt or milan.

"

Ispita wrote:

Real talk here. Even if it was routing problem you would have 40 ms instead of 20 ms (since the signal has to travel probably double the distance or more) but in this game it lags like hell which is obviously not routing issue.

When people talk about routing problems the general problem is that the traffic is going through congested or shit quality network nodes causing packet loss. It's not a matter of those packets taking longer to reach their target, they don't get received at all. There's a quite a good article explaining it here - it mostly mentions video calls but online games are something else that get fucked over badly by packet loss because the packets have to be processed in real time https://www.techtarget.com/searchnetworking/definition/packet-loss

Simple http traffic is much less affected by packet loss which is why you can have good upload/download speeds and still get screwed by packet loss while gaming.

"

Ispita wrote:

They could add more gateways that would spread the load that could potentially fix the issue. Add a new gateway to an eastern EU country maybe to romania or hungary or something so quarter of your eu playerbase won't be joined to the west EU places thus lifting a bunch of load from london/milan/amsterdam and frankfurt of course. Basically if you are from eastern europe or balkans your best option is frankfurt or milan.

That's fine in principle but do IBM have data centres in Romania or Hungary?

"

tsunamikun wrote:

[...]
The picture is pretty clear, there are hundreds of logs online, you could have done your research, but you didn't, how cute.
[...]

I don't care about your mythical "online" logs, there are none on this thread.

Proof or GTFO.

(And, advance warning to platform shills : your links will be ignored and you will be publicly shamed.)

"

Phrazz wrote:

If "bad routing" can cause instance "crashes" where you lose your portals, then OK; the problem might be my ISP for all I know. I just don't see why/how that's a routing issue. And everytime I get these sick-ass latency spikes, spiking into eternity, it ends with the map instance disappearing.

But I'm no computer wiz...

We're specifically NOT talking about instance crashes (unless there's a good reason this could be caused by routing ??), please don't confuse the issue.

"

Zrevnur wrote:

"

BlueTemplar85 wrote:

If the issue was exclusively on GGG's end, then everyone would have issues. But that's clearly not the case (my own experience with an EU server has been rock solid for the weeks I've been playing).

Unless you got some inside information or some other kind of information that I dont have: Not necessarily so. They may be (for example) distributing players according to some only-known-to-them-or-not-even-them scheme onto their servers for example. If they have limitations (obviously they have) then for performance optimization and such it could make sense to do that. If I would be responsible for that kind of tech the question would be expected cost of implementing a not-fully-random system vs expected benefit. There are a lot of variables available to use like crashes, load, bandwith, packet losses, instance size, expected instance duration, ...

I guess this is possible, but surely I would have gotten unlucky at some point ??
Unless GGG is prioritizing new players / recent supporters ?? (Kind of verging into conspiracy theory here...)

"

Zrevnur wrote:

"

BlueTemplar85 wrote:

As long as none of the complainers have provided proof - that OP already asked for, but is yet to be shown - that the issue is (more) on GGG's side than on theirs, it has to be assumed that the issue lies more with the complainers.

If not all is solely on GGGs side then there are multiple issues. Disappearing portals and rollbacks are GGG side obviously - neither players nor connection with players have control or access to that.

These too seem to be offtopic ?
(Again, unless there's a reasonable cause to believe that this is caused by routing ??)

"

Ispita wrote:

[...]
THe hosts they are renting runs dozens of services other games not just poe. If it was a big issue region wide don't you think it was not adressed by now? There are far more important and bigger things ran on the same hosts at the same time than poe.
[...]

I guess that you've just gotten lucky with ISPs so far ? Some of them can do really shitty routing to save money, even big ones !
Free (Iliad) has been pretty infamous for that for years, not sure whether they have fixed that issue yet where Paris to Paris was going through New York ?

"

When people talk about routing problems the general problem is that the traffic is going through congested or shit quality network nodes causing packet loss. It's not a matter of those packets taking longer to reach their target, they don't get received at all. There's a quite a good article explaining it here - it mostly mentions video calls but online games are something else that get fucked over badly by packet loss because the packets have to be processed in real time https://www.techtarget.com/searchnetworking/definition/packet-loss

Simple http traffic is much less affected by packet loss which is why you can have good upload/download speeds and still get screwed by packet loss while gaming.

this is not entirely true because other services would be impacted as well not just poe if the nodes are trash and trust me they are not likely to be. Capacities are way higher on the nodes it always comes down to the hosts and only because they want to save money not because hosts aren't capable of higher performance.

If you were around 15-20 years ago and you played online games they were running the same as today in terms of server to client performance but back then we had 2 core xeons not 100 core epycs and we had 10 gbit backbone connections to the hosts not terrabits like today. There is no reason to be any conection issue it is all about cheaping out on servers.

"

That's fine in principle but do IBM have data centres in Romania or Hungary?

Care to elaborate why it has to be IBM? There are plenty of other hosts that are just as good if not better. Also romania has like one of the best internet structure in EU hungary is close too. I'm sure they could find something good enough. If not there is still many other options like Serbia/Poland/Bulgaria in that region.

My point is a lot of players in eu that are resides in balkans/eastern europe has to join west servers because that is their best option. When you see a gateway germany does not mean only germans join there or something.

"

BlueTemplar85 wrote:

I don't care about your mythical "online" logs, there are none on this thread.

Proof or GTFO.

Go to Technical Support, that is the subforum for, DUMDUMDUM, technical issues. This is a thread in general discussion, why would people post troubleshooting information here when GGG supplies its own subforum for it and specifically ask and receive logs for winMTR there?

"

I guess this is possible, but surely I would have gotten unlucky at some point ??
Unless GGG is prioritizing new players / recent supporters ?? (Kind of verging into conspiracy theory here...)

I also played without problems for some weeks know, but yesterday I got those issues again. One sample doesn't make a statistic. Same for you. Especially as you haven't played that much.

"

These too seem to be offtopic ?
(Again, unless there's a reasonable cause to believe that this is caused by routing ??)

No, as heavy lag triggers this issue. It always begins with stuttering and a after a few seconds full on latency spike(s) as well as disconnect when reaching 100k+ ms.
I would guess the packet loss got high enough that the instance server flags it as a disconnect and rolls back the instance, or the instance itself crashed and force-disconnected the player, no way to know.