Kaapi Konsole: Running large evaluations

[AkhileshNegi]

Describe the bug
Uploading datasets with more than 50 rows returns a 504 Gateway Timeout error, even though the operation completes successfully in the
background. The same timeout behavior occurs when running evaluations with higher duplication factors (e.g., 5x).

To Reproduce

1. Go to Kaapi Konsole
2. Upload a dataset with more than 50 rows
3. Wait for the upload to process
4. See 504 Gateway Timeout error (despite operation completing in background)

While running evaluation:

1. Create/select a dataset with large number of rows
2. Run evaluation with duplication factor set to 5
3. See 504 timeout error

Expected behavior
Long-running operations should either Complete without timeout errors, OR Return immediately with a job ID and process asynchronously, showing progress/completion status in the UI

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

_{You can configure auto-planning by selecting labels in the issue_enrichment configuration.}

🔗 Similar Issues

Possible Duplicates

• ai-platform#460

Related Issues

• Kaapi Konsole: Implement UI on top of Kaapi APIs #458
• Evaluation: Optimize Upload Dataset to Langfuse Function to prevent timeout error #460
• Sudden surge in Kaapi responses API timing out #444
• Evaluation: Cron on staging #448
• Collections endpoints: Silent Failures Causes Incorrect 200 Responses on Error #238

👤 Suggested Assignees

• Prajna1999
• shijithkjayan
• AkhileshNegi
• nishika26

---

🧪 Issue enrichment is currently in open beta.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord or schedule a call!

[Prajna1999]

While this issue has been fixed thanks to #461, it still shows the same error (504 or 502 despite successful uploading of dataset) at about ~800 total dataset line items. Trigerred the same with APIs and they dont give 504 even at 1000 lines. I am guessing its the Netlify's proxy that preventing long running job beyond a ~60 seconds timeout but success at the kaapi-backend level. Once the console is migrated to Kaapi's EC2, will be able to control the proxy server more granularly to fix the issue.