Trouble Batch Inserting Records Using Langchain w/ Weaviate

Support
1

Description

I get the following error message running a batch job to load records into my Weaviate nodes.

Reason: local index "<INDEX_NAME>" not found: deadline exceeded for waiting for update: version got=0 want=64

After doing some digging I found this on my node:

{"got":0,"level":"debug","msg":"wait for update version","time":"2024-07-22T21:49:57Z","want":64}

Any ideas what could be causing this?

Server Setup Information

  • Weaviate Server Version: 1.25.8
  • Deployment Method: Docker on AWS ECS using Fargate
  • Multi Node? Number of Running Nodes: 2
  • Client Language and Version: Python 4.6.5
  • Multitenancy?: Yes
2

Oh, had to increase my timeout time for this. Here’s the config I ended up using:

weaviate.config.AdditionalConfig(
            timeout=weaviate.config.Timeout(init=30, query=60, insert=120)
        )
3

Hi!

Can you share the docker compose you have used?

Also, for multi node, it is best to have at least 3 nodes. :thinking:

4

I’m using Services and Task Definitions on AWS ECS so it’s a bit tricky to mimic a Docker compose. I’ll increase the nodes to 3 based on recommendation, but what is the benefit of doing that?

5

Hi!

It should show on logs, but if not properly configured, it may have a hard time defining the leader in the cluster:

However, not sure it would produce the symptoms you had, but wouldn’t work at all.

1 Like
6

Perfect, I’ll dig in and read.