mirror of
https://github.com/langgenius/dify-official-plugins.git
synced 2026-07-21 17:45:23 -04:00
Timeout plugin daemon vertex ai embedding #321
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @nurrochmanmuhammad on GitHub (May 28, 2025).
Originally assigned to: @JohnJyong on GitHub.
Self Checks
Dify version
1.4.1
Cloud or Self Hosted
Self Hosted (Docker)
Steps to reproduce
I am using vertex ai plugin 0.15
I have a document that contains some table in html format.
While I upload through web then it works for indexing, but when I add that file through API then indexing stuck, api keep retry until error timeout.
But when I try on cloud dify it works perfectly. Not sure what happened.
Anyone can help me on this.
✔️ Expected Behavior
Succeed indexing process
❌ Actual Behavior
Error timeout
@dosubot[bot] commented on GitHub (May 28, 2025):
Hi @nurrochmanmuhammad! I'm Dosu and I’m helping the dify team.
Timeout errors with the Vertex AI plugin in self-hosted Dify (Docker) are often caused by network configuration, insufficient timeout settings, or quota limits on the Vertex AI side. Since your indexing works via the web UI but not via API, and works on cloud Dify, this points to an environment/configuration issue in your self-hosted setup.
Here are targeted steps to resolve or diagnose the problem:
Check Vertex AI Quotas: If you see repeated retries and eventual timeouts, you may be hitting a quota or rate limit on your Google Cloud Vertex AI account. Check your Google Cloud console for quota errors, especially for online prediction requests. If you see quota exceeded errors, request a quota increase as described in the Vertex AI quotas documentation citation.
Increase Plugin Daemon Timeout: The default timeout for plugin execution may be too low for large or complex documents. Increase the timeout by setting
PLUGIN_MAX_EXECUTION_TIMEOUT(e.g., to 2400 seconds) in yourdocker-compose.yamlunder theplugin_daemonservice, and restart your Docker containers to apply the change citation.Adjust Network and Firewall Settings: Ensure your Docker host can reach external services without restrictions. Open necessary ports and check for firewall or proxy rules that might block outbound requests citation.
Tune HTTP Timeout Settings: In your
.envfile, increase values forHTTP_REQUEST_MAX_CONNECT_TIMEOUT,HTTP_REQUEST_MAX_READ_TIMEOUT, andHTTP_REQUEST_MAX_WRITE_TIMEOUTto allow longer waits for slow responses citation.Clear Plugin Data and Reinstall: If the plugin daemon is stuck or corrupted, clear plugin data from the database and plugin directories, restart the plugin daemon, and reinstall the Vertex AI plugin citation.
Check Logs for Specific Errors: Review your plugin daemon and Dify logs for messages like "Requests rate limit exceeded", "Worker was sent SIGKILL", or quota errors. These can pinpoint whether the issue is with Dify, Docker, or Vertex AI citation.
Update to Latest Plugin Version: Ensure you are using the latest official Vertex AI plugin, as previous timeout issues were resolved by upgrading the plugin citation.
If you try these steps and still encounter issues, please provide any error logs or quota messages you see, as they will help narrow down the root cause.
To reply, just mention my name @dosu.
How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other
@nurrochmanmuhammad commented on GitHub (May 28, 2025):
This doesnt work, the unique things only happen with that kind of file, table in html format and mix of markdown. I can consistently reproduce the same error here.
@crazywoola commented on GitHub (May 28, 2025):
Please provide some error logs, steps to reproduce this.
@nurrochmanmuhammad commented on GitHub (May 29, 2025):
steps:
the unique things happened, If I try to chunk the text per 1k char, then it works for entire doc, but if I give entire text then it stuck.
@nurrochmanmuhammad commented on GitHub (Jun 6, 2025):
Is there any update ?
Using openai plugin works fine, not with vertex ai plugin