pytorch / serve

Serve, optimize and scale PyTorch models in production
https://pytorch.org/serve/
Apache License 2.0
4.19k stars 857 forks source link

Netty Threads don't change even set explicitly in config.properties #1714

Closed Hegelim closed 2 years ago

Hegelim commented 2 years ago

https://github.com/pytorch/serve/issues/1700#issuecomment-1164746780

Any ideas why?

msaroufim commented 2 years ago

Both of those parameters are parameters to TorchServe and not to a specific model see this for an example https://github.com/pytorch/serve/blob/master/docker/config.properties#L1-L7

inference_address=http://0.0.0.0:8080
management_address=http://0.0.0.0:8081
metrics_address=http://0.0.0.0:8082
number_of_netty_threads=32
job_queue_size=1000
model_store=/home/model-server/model-store
workflow_store=/home/model-server/wf-store

And in your code you've shown them as model specific parameters

models={\
  "ABINet": {\
    "1.0": {\
         ...
        "number_of_netty_threads": 8,\
Hegelim commented 2 years ago

Ahh thanks! Finally figured it out!