Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Bug] Unable to reach Pebblo Server #226

Open
shreyas-damle opened this issue Feb 27, 2024 · 2 comments
Open

[Bug] Unable to reach Pebblo Server #226

shreyas-damle opened this issue Feb 27, 2024 · 2 comments
Labels
bug Something isn't working

Comments

@shreyas-damle
Copy link
Collaborator

shreyas-damle commented Feb 27, 2024

Description

When execute RAG app, we are getting error "unable to reach pebblo server." but it is generating report as expected on Pebblo Server.

Error message
$ python3 fin_corp_rag_app.py
Loading RAG documents ...
Unable to reach pebblo server.
Loaded 93 documents ...

Hydrating Vector DB ...
Finished hydrating Vector DB ...

Expected behavior
It should call Pebblo APIs and pdf report should get generated without any error.

Additional context
Pebblo server was healthy when this error occured.

System:

  • OS: Mac
  • GPU/CPU:
  • Pebblo version (commit or version number): 0.1.11
  • Langchain version: 0.1.9
  • DocumentStore:
  • Reader:
  • Retriever:
@shreyas-damle shreyas-damle added the bug Something isn't working label Feb 27, 2024
@shreyas-damle
Copy link
Collaborator Author

Observations:

  • When checked by enabling debug mode, langchain_community.document_loaders.pebblo is giving warning when it is timing out(timeout in code is 20 seconds). WARNING:langchain_community.document_loaders.pebblo:Unable to reach pebblo server.
  • The logs also say that DEBUG:urllib3.connectionpool:Starting new HTTP connection (1): localhost:8000, this means that the connection is established with Pebblo Server and server has accepted the request. Note that, "Unable to reach pebblo server" is custom message which is misleading here.
  • The exception is raised when timeout of 20 seconds is reached and in parallel, Pebblo Server is finished processing request with it's speed few seconds later if the payload is larger and generating report.

Issue is not on the Pebblo Server side, its the way connection timeout is handled on Pebblo Safe Loader side.

@rahul-trip
Copy link
Contributor

rahul-trip commented Feb 29, 2024

Below is the scenario when /loader/doc api server is busy classifying or generating report and pebblo langchain client waits for its response.

  • Reduced timeout to 5 seconds: reduced processing time by half, the plugin will not really wait for the response. (Earlier loading time was ~30 secs now it takes 15 secs to finish the loading(On my machine.))
  • Assumed pebblo-server and pebblo client relationship as async for now, hence changed warning logs to debug logs, this will suppress the unable to connect to pebblo server output and will now appear in debug mode.

@srics , can we reduce it even further? please share your thoughts.

PR: daxa-ai/langchain-rel-0.1.12#21

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

2 participants