Abnormal disconnect: Failed to connect to instance

Hi GGG Forum Tech Support,

Over the past few days I've started getting the error "Abnormal disconnect: Failed to connect to instance", frequently, when changing zones. This disconnection is only happening during an instance change. Seems more likely to happen when visiting a new instance.


Client Type: Steam
Gateway: Australia
NIC: Wired
ISP: Telstra, NBN


Interestingly enough from looking at the logs I can even be leaving an instance created on the same IP as I'm joining and still have this problem occur.

WinMTR below is for IP, 27.50.70.10 one of the instances I had just been disconnected from.

I've also had the problem for these IPs as well.

159.122.82.213:6112
27.50.70.10:6112
168.1.82.198:6112
168.1.82.197:6112
168.1.82.196:6112
168.1.82.205:6112
216.12.203.93:6112

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| home.gateway.home.gateway - 0 | 463 | 463 | 0 | 0 | 2 | 0 |
| No response from host - 100 | 93 | 0 | 0 | 0 | 0 | 0 |
| 144.130.208.210 - 1 | 459 | 458 | 4 | 4 | 10 | 4 |
|bundle-ether14.ken-core10.sydney.telstra.net -0 | 463 | 463 | 4 | 5 | 10 | 5 |
|bundle-ether1.ken-edge901.sydney.telstra.net -0 | 463 | 463 | 4 | 4 | 10 | 5 |
| equini2.lnk.telstra.net - 0 | 463 | 463 | 4 | 6 | 83 | 4 |
| xe-1-0-0.gw102.sy1.ap.equinix.com - 0 | 463 | 463 | 4 | 6 | 77 | 4 |
| ve1024.br-1-2.as45671.net.au - 0 | 463 | 463 | 4 | 5 | 10 | 5 |
|rtif-221.121.137.69-31.cs-1.as45671.net.au - 0 | 463 | 463 | 4 | 5 | 10 | 4 |
| vl2928.sw-1-1.as45671.net.au - 1 | 460 | 459 | 4 | 6 | 124 | 5 |
| sau-a49d0-or.servercontrol.com.au - 0 | 463 | 463 | 6 | 7 | 12 | 6 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider

Have you tried connecting to a different server? The lag time on those pings is a little high, but that doesn't mean you're getting disconnected or dropping packets. I do see a 1% packet loss on one of the final hops and perhaps your local neighborhood box, but that doesn't mean too much.

Can you run the WinMTR again with a delay of 0.1 seconds instead of 1 second between pings? That might give you a better picture of any real packet loss/disconnect.

Also, what technology is your ISP using? Coaxial broadband, DSL, satellite, or something else?

Tried switching out my router with the ISP supplied one to see if it would make a difference.

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| dsldevice.gateway - 15 | 129 | 110 | 0 | 0 | 1 | 0 |
| No response from host - 100 | 21 | 0 | 0 | 0 | 0 | 0 |
| 144.130.208.210 - 0 | 1054 | 1054 | 4 | 4 | 7 | 4 |
|bundle-ether14.ken-core10.sydney.telstra.net - 0 | 1054 | 1054 | 4 | 5 | 7 | 6 |
|bundle-ether12.win-core10.melbourne.telstra.net - 1 | 1007 | 1006 | 15 | 16 | 20 | 17 |
|bundle-ether1.win-edge902.melbourne.telstra.net - 0 | 1054 | 1054 | 15 | 15 | 18 | 15 |
| sof1522605.lnk.telstra.net - 0 | 1054 | 1054 | 15 | 15 | 33 | 16 |
| ae6.dar01.mel01.networklayer.com - 0 | 1054 | 1054 | 15 | 16 | 36 | 16 |
| po1.fcr01b.mel01.networklayer.com - 0 | 1054 | 1054 | 16 | 16 | 29 | 16 |
| c5.52.01a8.ip4.static.sl-reverse.com - 0 | 1054 | 1054 | 17 | 17 | 19 | 17 |
|________________________________________________|______|______|______|______|______|______|
WinMTR v0.92 GPL V2 by Appnor MSP - Fully Managed Hosting & Cloud Provider


Fibre to the premise, IPoE from the NTU to the router.

"

cipher_nemo wrote:

Can you run the WinMTR again with a delay of 0.1 seconds instead of 1 second between pings? That might give you a better picture of any real packet loss/disconnect.

i wouldn't recommend increasing the ping frequency, if other people follow that advice they would ddos the servers.
as a consequence the server would have to drop icmp packets to further prioritise game (tcp) packets, resulting in other players running a mtr to believe the game server being the problem because it drops icmp packets.

a traceroute is no reliable tool to diagnose a connection, it can just point to problematic (congested) packet routes or connection problems of your equipment.

"

crimsonfalcon wrote:

Fibre to the premise, IPoE from the NTU to the router.

There might be something wrong with fiber in your area.

I noticed this other thread related to what looks like the same type of connection issue. Just a hunch, no clue where this other person is...

https://www.pathofexile.com/forum/view-thread/1707341

And many other disconnect threads in this forum, more than usual.

Can you try troubleshooting from the assumption there's something wrong with your ISP? Other servers and sites, additional software for QoS of your connection.

"

crimsonfalcon wrote:

Tried switching out my router with the ISP supplied one to see if it would make a difference.

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| dsldevice.gateway - 15 | 129 | 110 | 0 | 0 | 1 | 0 |

your provider's router should not drop icmp packets. it points to a bad conection of your router (bad sync), try restarting it to see if the next sync brings better results.

otherwise i would take that other router, seemed to work not perfectly but fine.

"

vio wrote:

i wouldn't recommend increasing the ping frequency, if other people follow that advice they would ddos the servers.
as a consequence the server would have to drop icmp packets to further prioritise game tcp) packets, resulting in other players running a mtr to believe the game server being the problem because it drops icmp packets.

That won't happen. Network hardware has way more capacity than you think, not to mention plenty of protections from denial of service attacks. It takes a multitude of millions more to be considered such a threat to networks and access points (usually in the range of 400 to 600 Gb/s). One person setting their retry time from 1 second to 100ms isn't going to do anything, let alone even thousands of people doing that at the same exact time. Don't spread misinformation, please.

"

vio wrote:

"

crimsonfalcon wrote:

Tried switching out my router with the ISP supplied one to see if it would make a difference.

|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| dsldevice.gateway - 15 | 129 | 110 | 0 | 0 | 1 | 0 |

your provider's router should not drop icmp packets. it points to a bad conection of your router (bad sync), try restarting it to see if the next sync brings better results.

otherwise i would take that other router, seemed to work not perfectly but fine.

Considering I was sending 10 icmp packets per second to it it's more likely that the firewall is just rate limiting the ICMP packets, same problem doesn't happen if I send them at 1.0 second intervals.

"

cipher_nemo wrote:

"

crimsonfalcon wrote:

Fibre to the premise, IPoE from the NTU to the router.

There might be something wrong with fiber in your area.

I noticed this other thread related to what looks like the same type of connection issue. Just a hunch, no clue where this other person is...

https://www.pathofexile.com/forum/view-thread/1707341

And many other disconnect threads in this forum, more than usual.

Can you try troubleshooting from the assumption there's something wrong with your ISP? Other servers and sites, additional software for QoS of your connection.

That person was on the Frankfurt gateway unlikely that we would be using the same ISP as I'm in Australia.

There is every possibility that the problem is the provider but there certainly doesn't seem to be any issues with other sites or games I'm playing and there is an abnormally high amount of people having the same issue at vastly different locations.

"

crimsonfalcon wrote:

Considering I was sending 10 icmp packets per second to it it's more likely that the firewall is just rate limiting the ICMP packets, same problem doesn't happen if I send them at 1.0 second intervals.

That's very possible. Usually I see most networking hardware accept down to 100ms intervals, but different access points and networking hardware are owned and operated by a lot of different companies and institutions, so what they set it to will vary.

"

crimsonfalcon wrote:

That person was on the Frankfurt gateway unlikely that we would be using the same ISP as I'm in Australia.

There is every possibility that the problem is the provider but there certainly doesn't seem to be any issues with other sites or games I'm playing and there is an abnormally high amount of people having the same issue at vastly different locations.

If other servers and games are fine, we can go back to troubleshooting with your system then. The only other option is sending GGG a support request for the gateway/server you connect to with PoE. Have you tried other PoE servers?

"

cipher_nemo wrote:

Network hardware has way more capacity than you think, not to mention plenty of protections from denial of service attacks.

sorry, DoS or distributed DoS probably is the wrong term to use as it means attempts to bring down the servers which does not apply here.

for the negative effect i mentioned it's enough that the server has to limit serving icmp requests but even that is unlikely given how few winTMR reports we get here :-)

so please ignore my post above.