Flyte enables production-grade orchestration for machine learning workflows and data processing created to accelerate local workflows to production.

Flyte

image.png

Hey everyone
At large scale(hundreds of paralel tasks), we are getting 503 errors. what can be done?

I'll ask our devops tommorow <@U05QE97KFS4>. can you give me some context on why you are asking that?

yes, the reason is, single binary deployment is a little low powered

you can try increasing the memory / cpu for the deployment

this will help, but if it does not, you will have to deploy the flyte-core helm chart

Thanks <@UNZB4NW3S>, I can confirm that we are in fact using flyte-binary, will implement your suggestions :pray:

<@UNZB4NW3S> Is there an example of how to increase resources in flyte-binary? I'm guessing it's the cluster_resources section within configuration.inline but the <https://docs.flyte.org/en/latest/deployment/configuration/generated/flyteadmin_config.html#section-cluster-resources|documentation> does not have any more specifics

NVM, found it, here's what I updated to:
```deployment:
  image:
    tag: "v1.10.6"
  resources:
    requests:
      cpu: 0.5
      memory: 20Gi
    limits:
      cpu: 2
      memory: 32Gi```

Can you please increase the CPU - 0.5 is nothing. Atleast give it 2. It does a lot of v things right