Self Hosting 5 min Timeout?

Support
1

Hi,

I am using a standard docker setup, for text, that will store large scripts of text in an array. I have about 8Gb Ram 2Vcpu and 1/8th of an A16 (2GB GPU Ram). It seems to run okay for a few objects being uploaded but pretty much timesout like taking 5 minutes per update…

I probably have 1000 Objects and scripts for only 100.

Is there a way to understand better the possible throughput for data? I don’t think its a small machine, but perhaps I am wrong? Why would this slowly stop working? I attempted rebooting and bring docker down and back up, but no help. Is there a better way to log issues to find the problem?

version: '3.4'
services:
  weaviate:
    command:
    - --host
    - 0.0.0.0
    - --port
    - '8080'
    - --scheme
    - http
    image: semitechnologies/weaviate:1.20.3
    ports:
    - 8080:8080
    restart: unless-stopped
    volumes:
      - /mnt/data-storage/weaviate:/var/lib/weaviate
    deploy:
      resources:
        reservations:
          devices:
            - capabilities: [gpu]
    environment:
      TRANSFORMERS_INFERENCE_API: 'http://t2v-transformers:8080'
      SPELLCHECK_INFERENCE_API: 'http://text-spellcheck:8080'
      QUERY_DEFAULTS_LIMIT: 25
      AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: 'true'
      PERSISTENCE_DATA_PATH: '/var/lib/weaviate'
      DEFAULT_VECTORIZER_MODULE: 'text2vec-transformers'
      ENABLE_MODULES: 'text2vec-transformers,text-spellcheck'
      CLUSTER_HOSTNAME: 'node1'
      LOG_LEVEL: 'debug'
  t2v-transformers:
    image: semitechnologies/transformers-inference:sentence-transformers-multi-qa-MiniLM-L6-cos-v1
    environment:
      ENABLE_CUDA: '1'
      NVIDIA_VISIBLE_DEVICES: 'all'
    deploy:
      resources:
        reservations:
          devices:
          - capabilities: 
            - 'gpu'
2

hi @msj242, Welcome to our Community :slight_smile:

have you seen this doc?

Even though, according to that doc, your usecase is “small use cases (less than 1M objects)”, it might give you an insight on what can be the bottleneck in your scenario.

Can you see any outstanding logs? Also, what version of Weaviate are you in 1.20.3

3

I have tidied up my code for uploading and I realized I was hitting the server too hard. I had thought I was sleeping whilst due to my threading I was not and I was destroying it myself creating massive data bottlenecks due to some race conditions. My bad.

4

Glad you solved it!

Thanks for sharing, this will certainly help many others to come. :slight_smile: