BUG REPORT 3013 2014 3001 (backend)

Issue Description (Required):

I am experiencing recurring backend connectivity failures in Warhammer 40,000: Darktide. The game frequently fails to connect to backend services or disconnects during matchmaking or mission loading, resulting in error codes 3001, 3013, and occasionally 2014.

According to console.log, the issue is caused by backend connection and resolving timeouts when attempting to connect via gRPC to bsp-im-prod.atoma.cloud:443. The logs consistently show resolving and handshake timeouts rather than client-side crashes or authentication failures.

The issue occurs intermittently but frequently enough to prevent stable gameplay.

Recurring backend connection timeouts (Error codes 3001 / 3013 / 2014) – gRPC resolving timeout to atoma.cloud


Description

I am experiencing recurring and persistent connection failures in Warhammer 40,000: Darktide.
The game frequently fails to connect to backend services or disconnects during matchmaking or mission loading, resulting in error codes 3001, 3013, and occasionally 2014.

The issue occurs intermittently but has become frequent enough to prevent normal gameplay.


Error Codes Observed

  • 3001
  • 3013
  • 2014

Log Evidence

From console.log:

[Backend] Resolving timed out after 5003 milliseconds
Code: 1000
Description: Resolving timed out after 5003 milliseconds
[GRPCManager] Connecting to grpc address bsp-im-prod.atoma.cloud:443

Additional backend-related errors observed:

  • failed_handshake_timeout
  • pong_timeout
  • failed_found_no_lobby
  • nonexisting_channel

These indicate backend connectivity and handshake failures rather than client-side crashes.


Environment

  • OS: Windows 11
  • Platform: Steam
  • Network: Home broadband (Wi-Fi router)
  • Region: Russia

Troubleshooting Already Attempted

The following steps were tested extensively and did not resolve the issue:

  • Changing DNS providers (multiple public DNS)
  • Flushing DNS cache and resetting network stack
  • Using VPNs with different servers and protocols (including encrypted tunnels)
  • Disabling VPN entirely
  • Restarting router and PC
  • Reinstalling the game
  • Switching between Wi-Fi and Ethernet (when available)

The problem persists across all of the above scenarios.


Additional Notes

  • The timeout occurs specifically during backend resolution / connection to atoma.cloud infrastructure via gRPC over port 443.
  • This strongly suggests a network routing, DPI, or regional connectivity issue rather than a local configuration problem.
  • Similar reports are being posted by other players, particularly from Russia, but not limited exclusively to this region.

Attachments

  • console.log
  • darktide_launcher.log

Request

Please investigate potential backend routing, gRPC connectivity, or regional network compatibility issues affecting connections to atoma.cloud, particularly for players in Russia.

If additional diagnostics or traces are required, I am willing to provide them.

Thank you for your time and support.

Attempted Solutions (Optional):

Changed DNS providers (multiple public DNS services)

Flushed DNS cache and reset Windows network stack

Restarted PC and router

Tested with VPN enabled and disabled (multiple servers and protocols)

Switched between different VPN providers and encryption methods

Reinstalled the game

Tested different network configurations (Wi-Fi / Ethernet when available)

None of the above resolved the issue.

Platform (Required):

PC - Steam

[PC] PC Specifications (Optional):

Operating System: Windows 11

Network: Home broadband connection via Wi-Fi router

Platform: Steam

Upload Supporting Evidence (Optional):

console-2025-12-13-17.23.08-cd05b291-c462-4679-81e1-3c359f9fbfa2.log (331 KB)

Additional Technical Context (Network / Backend)

Based on repeated testing and log analysis, the connection issues (error codes 2001 / 2014 / 3013 / 4001) appear to be triggered by a combination of network-level packet deformation upstream and backend sensitivity to short-lived traffic bursts, rather than by client misconfiguration.

Observed Behavior

  • Client logs consistently show:

    • Resolving timed out

    • failed_handshake_timeout

    • pong_timeout

    • gRPC connection attempts to bsp-im-prod.atoma.cloud:443

  • Average latency and throughput appear normal, but connection failures occur during:

    • initial backend handshake

    • matchmaking / lobby join

    • keepalive ping-pong phases

  • The problem is more frequent in Russia but also reported in other regions, suggesting amplification rather than a strictly regional outage.

Key Observation

A combination of DPI-avoidance tooling (e.g. packet-level TLS fragmentation / ClientHello splitting) and traffic tunneling via Cloudflare WARP significantly reduces or fully eliminates the issue for many affected users.

This strongly suggests that:

  • The failures are not caused by slow average bandwidth

  • But by microbursts, packet reordering, or short packet loss events

  • Which are sufficient to disrupt gRPC / HTTP2 streams and backend keepalive logic

Likely Contributing Factors

  1. Upstream DPI / traffic inspection

    • Introduces packet fragmentation, delay, or reordering

    • Does not fully block traffic but deforms it

  2. gRPC over HTTP/2 sensitivity

    • High PPS patterns

    • Tight keepalive / ping-pong timing

    • Low tolerance to packet loss or reordering

  3. AWS ENA limits and burst behavior

    • Potential bw_out_allowance_exceeded or pps_allowance_exceeded

    • Silent packet drops during short traffic spikes

    • Effects amplified on long-haul routes with higher RTT variance

Why DPI Workarounds Appear to “Fix” the Issue

DPI-avoidance tools do not repair backend infrastructure; instead, they:

  • Prevent upstream packet deformation

  • Reduce retransmit amplification

  • Smooth traffic bursts before reaching AWS
    As a result, backend services receive traffic in a form they can handle reliably.

Suggested Areas for Investigation (Backend Side)

  • CloudWatch metrics:

    • bw_in_allowance_exceeded

    • bw_out_allowance_exceeded

    • pps_allowance_exceeded

    • conntrack_allowance_available

  • gRPC configuration review:

    • keepalive intervals and timeouts

    • tolerance for transient packet loss

    • retry / backoff behavior

  • Instance sizing and network performance class

  • Traffic shaping or buffering between matchmaking and mission instances

Summary

This appears to be a cross-layer interaction issue, where upstream network deformation exposes fragility in backend networking assumptions. Improving tolerance to packet reordering, microbursts, and transient loss would likely reduce or eliminate these connection failures without requiring client-side workarounds.