Fix bug between Retriever, TEI-embedding and VectorDB

[zhlsunshine]

Description

There are dependencies among Retriever, TEI-embedding and VectorDB, the Retriever can get normal response if both EI-embedding and VectorDB are really up and running well. However, the retriever would get error and this kind of error is always here if the EI-embedding or VectorDB are not available, and I encounter this situation in my gaudi node (Not the first time to deploy). Please refer to log files in attach for more detail.

Fix solution: I add the check_embedding_endpoint to make sure that the embedding endpoint is available, then begin to retrieve the data from VectorDB via TEI-embedding. And the problem is fixed, the Retriever log is shown as below:

Error log files: retriever-svc.log redis-vector-db.log

Issues

n/a

Type of change

List the type of change like below. Please delete options that are not relevant.

• [x] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds new functionality)
• [ ] Breaking change (fix or feature that would break existing design and interface)

Dependencies

n/a

Tests

Describe the tests that you ran to verify your changes.

[zhlsunshine]

We can not find this issue on Xeon, I think the reason is that Xeon is slower than Gaudi to embed the pdf file into Redis vectorDB in retriever. And it seems that all pods are always running.