Skip to content

Latest commit

 

History

History
40 lines (27 loc) · 1.46 KB

RUN_GCP.md

File metadata and controls

40 lines (27 loc) · 1.46 KB
Raw

Run the project using GCP

differer runs on App Engine Standard environment in the smallest free tier allowed instance and configured to shut down itself if no requests arrive in 1 hour. This is my current app.yaml file:

runtime: go113
instance_class: F1

main: github.com/jimen0/differer/cmd/differer

default_expiration: 1h

env_variables:
  DIFFERER_CONFIG: "config.yaml"

You can deploy it as easily as gcloud app deploy . once you configure gcloud.

Each runner is a Cloud Run instance configured to be as small as possible but allowing them to process a decent number of concurrent requests:

gcloud run deploy python-parseurl \
    --image=gcr.io/REDACTED/python-parseurl:latest \
    --concurrency=60 \
    --platform=managed \
    --memory=128Mi \
    --max-instances=4 \
    --timeout=10 \
    --region=europe-west1

NOTE: Please note that this deployment will allow unauthenticated HTTP requests! You will have to tweak the configuration a bit if you want to protect your environment a bit more and only allow authenticated requests. No worries, Google's documentation is awesome.

There's no need to use Cloud Run. As long as your runners can receive the jobs over HTTP, they can run anywhere you want.

Diagram

Here's a simplified diagram of the GCP architecture. Please, remember to set the rules in App Engine and Cloud Run so only you can trigger the parsers!

GCP diagram