MetaHuman Auto-Rigging Face Mesh fails with 'Server Error' after 300s timeout — Always stuck at 75% — UE 5.7

Development Character & Animation
,
1

I’m stuck. I’ve tried everything on my end with no luck. Could use some help. Here are the details:

Engine Version: UE 5.7 (Release-5.7-CL-51494982)
OS: Windows 10.0.19041.5915 (64-bit)
Network: Personal home network, proxy enabled (tested with Dallas TX and Los Angeles CA nodes), also tested without proxy

Problem Description

Auto-Rigging Face Mesh consistently fails after 300 seconds timeout at exactly 75% progress. The Auto-Rigging job is successfully created on the server (Job ID generated), but the polling request (GET /v1/autorigs/{jobId}/status?waitSeconds=10) never returns any result — always times out and returns Server Error.

This issue has been reproduced with multiple different MetaHuman presets and configurations (see Test Matrix below).

Test Matrix

Test # Preset Modifications Rig Type Proxy Result Job ID
1 Aera (female) Body + Face Blend shapes adjusted Full Rig No proxy :cross_mark: Same failure 3ef04a74-33b2-4983-8d00-e7ad3c21d6c0
2 Bo (default, no modifications) None Full Rig No proxy :cross_mark: Same failure (This time it fails before creating Jobid)
3 All above — — Dallas, TX node :cross_mark: Same failure —
4 All above — — Los Angeles, CA node :cross_mark: Same failure —

Conclusion: Issue is NOT caused by custom mesh modifications or preset choice. Both the modified Aera and the default Bo preset exhibit the exact same failure at 75% progress. This definitively rules out user mesh issues.

Debug #1 — Domain Connectivity Test (All Passed)

Method: PowerShell Invoke-WebRequest -Method Head -TimeoutSec 10

URL Result Notes
https://quixel.com :white_check_mark: 200 OK Normal
https://epicgames.com 403 Forbidden Cloudflare anti-bot, network is fine
https://dev.epicgames.com 403 Forbidden Cloudflare anti-bot, network is fine
https://s3.us-east-1.amazonaws.com 405 Method Not Allowed S3 root returns 405 for HEAD, normal
https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com 404 Not Found GET returns 404, DNS resolves, service exists 
PS> Invoke-WebRequest -Uri "https://quixel.com" -Method Head -TimeoutSec 10 -UseBasicParsing
StatusCode        : 200
StatusDescription : OK

PS> Invoke-WebRequest -Uri "https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com" -Method Head -TimeoutSec 10 -UseBasicParsing
The remote server returned an error: (404) Not Found.    ← Server responded, not a timeout

Conclusion: All domains resolve correctly, HTTPS 443 is reachable, no DNS pollution, no firewall blocking.

Debug #2 — POST Request Test (Proves API Is Functional)

Method: PowerShell Invoke-WebRequest -Method POST -TimeoutSec 15

PS> Invoke-WebRequest -Uri "https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com/v1/autorigs" `
    -Method POST `
    -Headers @{"Content-Type"="application/json"} `
    -Body '{"meshId":"test","rigType":"Full"}' `
    -TimeoutSec 15

The remote server returned an error: (401) Unauthorized.    ← Fast response (<1s), NOT a timeout

Analysis:

  • Server responds to POST requests normally, not a network timeout

  • 401 means the API endpoint exists and processes requests correctly

  • 401 is returned because no valid EOS Auth Token was provided in the test

Conclusion: The Auto-Rigging service API is fully functional and responding.

Debug #3 — UE Editor Log (Full Request Chain)

[13:26:10] LogMetaHumanCharacterEditor: Display: Requesting Auto-Rigging for MainMetaHumanCharacter. Rig Type: Joints and Blend Shapes. Report Progress: true. Blocking: false
[13:26:10] LogMetaHumanCharacterEditor: Display: Auto-Rigging Face Mesh
[13:26:11] LogEOSSDK: LogEOS: [Boot] EOSSDK Version 1.19.0.3-49960398 booting
[13:26:11] LogEOSSDK: LogEOS: [Boot] Platform Configuration win64-dynamic-x86_64
[13:26:12] LogEOSSDK: LogEOSAuth: UpdateUserAuthToken: User ClientId=xyz...lQa AccountId=2bb...c43
           Access[Expires: 2026.04.14-15.26:09 Remaining: 7200.83]
           Refresh[Expires: 2026-10-11T07:46:09.556Z Remaining: 15531600.56]
           State: Valid
[13:26:12] LogEOSSDK: LogEOSAuth: UserAuthGenerated: Received FUserAuthToken
[13:26:12] LogMetaHumanAuth: Display: User name is DexterH
[13:26:12] LogEOSSDK: LogEOSMessaging: Successfully connected to Stomp.
[13:26:16] LogHttp: Warning: HTTP request timed out after 3.00 seconds URL=https://www.google.com/generate_204
           ... (multiple 3s timeouts for google.com — unrelated)
[13:31:50] LogHttp: Warning: HTTP request timed out after 300.00 seconds
           URL=https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com/v1/autorigs/3ef04a74-33b2-4983-8d00-e7ad3c21d6c0?waitSeconds=10
[13:31:50] LogMetaHumanCharacterEditor: Error: Auto-Rigging of Face failed with code 'Server Error'

Auth Token Verification:

  • Token expires at 15:26, current time is 13:26, Token is fully valid

  • EOS login: EOS_Success

  • Stomp messaging service: connected successfully

Note on 75% Progress: The Auto-Rigging process consistently gets stuck at 75% before the 300s timeout occurs, suggesting the server-side processing hangs specifically during the face mesh rigging phase.

Conclusion:

  1. Auth is completely normal, Token is valid and not expired

  2. UE established WebSocket connection to Epic services (Stomp)

  3. Job ID 3ef04a74-33b2-4983-8d00-e7ad3c21d6c0 was successfully created

  4. The issue is NOT on the client side — the task is created but never processed, polling never receives a result

Debug #4 — Windows Firewall Verification (No Blocking)

Windows Firewall rules confirmed no blocking. Manual curl -I test:

$ curl -I https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com
HTTP/1.1 404 Not Found
Server: awselb/2.0
← TLS handshake successful, server responding normally

Debug #5 — Live API Connectivity Test (2026-04-14 22:11:54)

All infrastructure endpoints respond correctly and quickly (264ms–3612ms).
No timeouts, no DNS failures, no firewall drops.
The 401 responses are expected (no client auth token in test).
The issue is 100% server-side: Auto-Rigging jobs are created but never processed.


================================================================================
MetaHuman Auto-Rigging — Live API Connectivity Test
Time: 2026-04-14 22:11:54
================================================================================

[TEST 1] POST /v1/autorigs (Create Auto-Rigging Job)
  URL: https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com/v1/autorigs
  Result: 401 Unauthorized
  Response Time: 1299ms ✅ (NOT a timeout — server responded quickly)
  X-Epic-Correlation-Id: d38dc939-865d-4e61-afde-ccb0363b679b
  X-Epic-Error-Code: 1002 (errors.com.epicgames.unauthorized)
  Conclusion: API endpoint is alive and correctly enforcing authentication

[TEST 2] GET /v1/autorigs/{jobId} (Get Job Details)
  URL: https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com/v1/autorigs/3ef04a74-33b2-4983-8d00-e7ad3c21d6c0
  Result: 401 Unauthorized
  Response Time: 264ms ✅ (NOT a timeout)
  X-Epic-Correlation-Id: f1e9a478-4ca5-4620-b992-7ae798c32a13
  Conclusion: Job ID endpoint exists and correctly enforces auth

[TEST 3] GET /v1/autorigs/{jobId}/status (Poll Job Status)
  URL: https://mh-uemhc-autorig-service.eeeb.live.use1a.on.epicgames.com/v1/autorigs/3ef04a74-33b2-4983-8d00-e7ad3c21d6c0/status
  Result: 401 Unauthorized
  Response Time: 263ms ✅ (NOT a timeout)
  X-Epic-Correlation-Id: 0506e280-8fe3-427e-a7ab-9140932d0bcf
  Conclusion: Status polling endpoint is alive and auth-enforced

[TEST 4] Epic OIDC Provider (Epic Auth Infrastructure)
  URL: https://api.epicgames.dev/epic/oauth/v1/.well-known/openid-configuration
  Result: 200 OK ✅
  Response Time: 1112ms
  Conclusion: Epic Auth infrastructure is fully operational

[TEST 5] Quixel CDN
  URL: https://www.quixel.com
  Result: 200 OK ✅
  Response Time: 3612ms
  Conclusion: Quixel/Megascans CDN is fully operational

[TEST 6] AWS S3 (us-east-1)
  URL: https://s3.us-east-1.amazonaws.com
  Result: 405 Method Not Allowed ✅ (Expected — S3 root only allows GET)
  Response Time: 1216ms
  x-amz-request-id: 0VGJGJ15DSF41ZCS
  Server: AmazonS3
  Conclusion: S3 storage endpoint is fully operational

================================================================================
SUMMARY: All infrastructure endpoints respond correctly and quickly (264ms–3612ms).
No timeouts, no DNS failures, no firewall drops.
The 401 responses are expected (no client auth token was provided in the test).
The issue is 100% server-side: Auto-Rigging jobs are created but never processed.
================================================================================

Workarounds Tried (All Failed)

# Method Result
1 Restart UE Editor :cross_mark: No effect
2 Clear EpicGamesLauncher Manifest cache :cross_mark: No effect
3 Enable proxy — Dallas, TX node :cross_mark: No effect
4 Enable proxy — Los Angeles, CA node :cross_mark: No effect
5 Disable proxy, direct connection :cross_mark: No effect
6 Re-login Epic account :cross_mark: No effect
7 Retry at different times of day (multiple attempts) :cross_mark: No effect
8 Different Preset (Aera with blend modifications) :cross_mark: Same failure
9 Different Preset (Bo, default, no modifications) :cross_mark: Same failure

Related Forum Threads

Link Status
https://forums.unrealengine.com/t/metahuman-auto-rigging-stuck-at-75-fails-with-error-ue-5-7/2709617 No official reply
https://forums.unrealengine.com/t/metahumans-error-auto-rigging-of-face-failed-with-code-server-error/2689790 No official reply
https://forums.unrealengine.com/t/server-error-preventing-the-metahuman-rig-from-being-imported-from-epics-server/2660563 Official confirmed firewall issue (enterprise users only)

Root Cause Analysis

All client-side debug evidence (network connectivity, API responses, valid Auth Token) conclusively shows that the issue is NOT on the client side.

The Auto-Rigging Job was successfully created, but the server-side processing queue appears to be stalled — the job is never consumed or processed. Critically, this failure occurs:

  • At exactly 75% progress every single time

  • With completely default presets (Bo, no modifications) — ruling out user mesh issues

  • With and without proxy — ruling out routing issues

  • Across multiple proxy nodes — ruling out specific node issues

Perhaps it would be helpful to check with the backend team to investigate the server-side logs for Job ID 3ef04a74-33b2-4983-8d00-e7ad3c21d6c0. This can confirm if the issue is a queue blockage or a server-side exception during the face mesh rigging phase.

2

Please help!