Vectara search works but cURL times out

Vectara Platform Q&A
1

This search works fine in the Vectara web app but using the cURL generated by clicking on Inspect does not and I get this reponse: {"responseSet":[{"response":[], "status":[], "document":[], "generated":[], "summary":[], "futureId":1}], "status":[{"code":"QRY__TIMEOUT", "statusDetail":"Operation timed out. Results are partial.", "cause":null}], "metrics":null}

cURL request

-H "Authorization: [Bearer XXX]" \
-H "customer-id: 3279618736" \
https://api.vectara.io:443/v1/query \
-d @- <<END;
{
  "query": [
    {
      "query": "Ongoing increases in the target range will be appropriate",
      "queryContext": "",
      "start": 0,
      "numResults": 10,
      "contextConfig": {
        "charsBefore": 0,
        "charsAfter": 0,
        "sentencesBefore": 2,
        "sentencesAfter": 2,
        "startTag": "%START_SNIPPET%",
        "endTag": "%END_SNIPPET%"
      },
      "corpusKey": [
        {
          "customerId": 3279618736,
          "corpusId": 2,
          "semantics": 0,
          "metadataFilter": "doc.country_iso_2='us' and doc.date>='2022-01-01' and doc.date<='2023-09-18'",
          "lexicalInterpolationConfig": {
            "lambda": 1
          },
          "dim": []
        }
      ],
      "summary": []
    }
  ]
}
END```
2

Hi,

You can copy the correct cURL as per your environment by clicking on the Inspect button.

image
image1368×823 124 KB

3

I tried that and the cURL does not work. I get the following:

{"responseSet":[{"response":[], "status":[], "document":[], "generated":[], "summary":[], "futureId":1}], "status":[{"code":"QRY__TIMEOUT", "statusDetail":"Operation timed out. Results are partial.", "cause":null}], "metrics":null}

Can you try replicating the example?

4

Can you please share your customer id and corpus id for which you are getting this? That will help us diagnose the problem better.

5

They’re in the cURL request I posted above.

customerId: 3279618736
corpusId: 2

6

Can you try with a timeout in the header like the following?

-H "grpc-timeout: 30S"

Also, I notice that you have set lambda to 1 which indicates that you are looking for only keyword-based search results. Keyword-based search results are generally slower than neural search. We are working on improving those.

From the Console:

Lambda ranges from 0 to 1. At the extremes, 0 will execute a search based solely on neural search and 1 will execute a search based solely on keyword-based search. Values in the range of 0.01 to 0.2 seem to work well for many cases.

7

Thanks, Aamir. Adding the timeout header solves things for now.