Incorrect routing to Singapore gateway – 213.* pool mis-routed via Europe for Asian Server

Hello GGG Support,

I would like to report a persistent routing issue with the Path of Exile Singapore gateway. This affects multiple players in India on different ISPs (Excitel, ACT, Airtel), and appears related to how one of your IP pools is being advertised via BGP.

Summary of the Problem

The Singapore gateway assigns players to two different server IP pools:

172.x.x.x pool – works correctly
• Route: India → Singapore
• Latency: ~85–100 ms stable
• No packet loss
• Consistent performance

213.x.x.x pool – incorrectly routed
• Route: India → Europe (Paris / Frankfurt) → Tokyo/US → Singapore
• Latency: 180–300 ms
• Jitter spikes and occasional packet loss
• Reloading the area usually reconnects us to a 172.* instance and latency returns to normal immediately

And rarely 161.x.x.x, which is also laggy and has incorrect route.

This behaviour has been tested repeatedly and is identical across different systems and ISPs. This strongly suggests the 213. subnet is being advertised from European peerings*, rather than originating from your Singapore POP.


Evidence Collected

• Multiple WinMTR logs
• Multiple tracert logs
• Tests from different ISPs and machines
• Real-time tests during instance switching
• Consistent correlation between pool and routing behaviour

The 172.* pool always routes directly from India to Singapore (AS6453 → NTT/i3D/SoftLayer), resulting in correct ~90 ms latency.
The 213.* pool regularly routes via Europe, causing doubled latency.

Recently, my ISP has re-routed its connection for 213. subnet because no matter what and how they tried, they could not actually make the data route travel to Singapore from Mumbai, as it is a European subnet, which is why it directs the connection to Paris or other EU servers. After re-routing the connection from their end, my game can not connect to 213. subnet anymore, but can do so with 172. subnet. ISP has made the same route for both the subnets, which is why 172 can connect to the server instance, while 213 cannot, as it is the wrong path taken.

When the game connects to one of the pool that 172, it connects right away and has stable ping, but when the game tries to connect to 213 subnet, it connot and gives out "failed to connect to instance" message and takes 4-5 tries for it to refresh the instance and connects to 172 subnet.


Likely Cause

A BGP origin/announcement issue involving the 213.* subnet.
It appears to be announced from OVH/NTT/LeaseWeb/i3D Europe instead of your Singapore location, which forces Indian ISPs to take a longer European path.

1. The 213. prefix appears to be advertised from Europe instead of the Singapore POP*

This is the strongest and most consistent finding in all traces.

When connecting to a 213.* instance, the AS-path selected by Indian ISPs includes European transit (Paris / Frankfurt) before eventually reaching Singapore.
This is consistent with a prefix that is:

> originated in Europe,

> or primarily announced by a European POP,

> or advertised into upstream peers from a European route reflector,

> or lacking a Singapore-region announcement.

Because BGP selects routes based on policy and AS-path length (not geography), Indian ISPs then prefer this European path, even though it creates a large geographic detour.

This behaviour explains the stable high latency (180–300 ms), the predictable detour, and the fact that returning to a 172.* instance immediately fixes the issue.

2. The 172. prefix is correctly announced from Asian/Singapore POPs*

This prefix consistently routes:
India → Mumbai → Singapore,
with a stable ~75–105 ms RTT and no packet loss.

This proves:

> Indian ISP paths to your Singapore infrastructure are healthy,

> there is no systemic ISP problem,

> and proper low-latency routing is already possible.

Only the 213.* announcement is problematic.

3. Additional possible contributing factors

> These are secondary but consistent with the symptoms:

> AS-path prepending on the Singapore-origin announcement for 213.*

> Route community tags causing the SG-origin to be deprioritised

> Anycast inconsistencies, where 213.* is only announced in EU POPs

> Transit congestion on EU legs used by the 213.* path

> Load balancing sending India → EU → APAC due to misconfigured regional policies

Why this is not an ISP or user-side issue

> Multiple unrelated ISPs in India (Excitel, ACT, Airtel) all take the same incorrect path for 213.*.

> The local hops and backbone segments up to SG for 172.* are stable and clean.

> Routing instantly fixes itself when the instance uses a 172.* IP.

> Packet loss and jitter only occur on the 213.* path, not the user’s local path.

> The problematic route includes EU transit, which no Indian ISP would choose unless BGP advertisements direct them there.


Request

Please escalate this to your network engineering team or your hosting provider (SoftLayer / i3D / NTT / OVH).
Kindly request:

• That the 213. pool used for Singapore* be announced from the Singapore POP, or
• That routing policies be adjusted for Asian clients, or
• That the 213.* pool be temporarily removed from the Singapore gateway rotation for Indian traffic.

The routing on the 172.* pool proves that stable and correct paths are fully possible.

Attachments Provided

• WinMTR – 172.* pool (good route)
• Traceroute logs
• Timestamped tests
• Client.txt excerpt

Thank you for your time. I am available to run additional tests if required. But cannot run tests for 213. subnet anymore, as that pool is not accessible to me anymore because it was causing issues, and the game can not connect to it because of its wrong data route.

Regards,
Sid
[Zepxxism#7411]
Region: India/Asian Server

---------------------------------------------------

172 Subnet Tracert



213 Subnet Tracert


172 Subnet WinMTR log
https://pastebin.com/n7Vrk0wq

Previous Distrorted Route WinMTR log
https://pastebin.com/sgC5mNrn

Latest Client.txt
https://pastebin.com/pbkFTjBw

Latency Chart for 172 Subnet


Latency Chart for 213 Subnet
I cannot share such evidence as I have not recorded it before. After ISP has put both subnets on the same route, I cannot connect to the 213 subnet and get a "failed to connect to instance" message.

-------------------------------------------------

1. Your Path of Exile Client - Steam

2. Your Computer -
> CPU - AMD 9950X3D
> GPU - RTX 4090 Aorus Waterforce Xtreme
> Motherboard - Asus x670E Crosshair Hero
> BIOS - 2904
> SSD - Samsung 990 PRO 4TB x2
> PSU - Corsair HX1500i
> RAM - Corsair Dominator Titanium 6600Mhz 2x16GB

3. Your Connection to the Internet - Ethernet Cable

4. Public Internet Connectivity - Yes

5. The Path of Exile Servers - Singapore

It has been almost a day since this post.
I am requesting GGG support to acknowledge this post and issue, and at least provide a reply.

+1 I have issues with this specific server as well

https://pastebin.com/a5jqRLDz

https://www.pathofexile.com/forum/view-thread/3425526

Think the issue is your ISP's problem.

The issue is likely cause by a connection issue around the fourth to fifth hop and is bad enough that you are being routed around, GGG at best can only notify your ISP, they have no power over them...YOU have more say than GGG in this matter.