search

SciDAS / nextflow-api

MIT License
31 stars 5 forks source link

remove old deployment files #30

Closed cbmckni closed 4 years ago

cbmckni commented 4 years ago

Removed old deployment files and updated helm files and docs. Works on PRP and GCP.

bentsherman commented 4 years ago

@cbmckni Can you add helm variables for the name and hostname in helm/values.yaml. Ideally those things shouldn't be hardcoded to a particular namespace.

Also can we go ahead and bump the resource request/limit up to 1 CPU and 4GB of memory for the web server pod?

cbmckni commented 4 years ago

@bentsherman I increased the resource requests.

As for your other points:

  1. Do you mean the hostname the Ingress uses? You can configure hostname in values.yaml:
# Ingress control settings
Ingress:
  # If true, use ingress control.
  # Otherwise, generic LoadBalancer networking will be used, 
  # and the other settings in this section will be ignored.
  Enabled: true
  # The subdomain to associate with this service.
  # This will result in a FQDN like {subdomain}.{cluster}.slateci.net
  Host: nf-api-teach-compbio.nautilus.optiputer.net
  # The class of the ingress controller to use. 
  # For SLATE this should be 'slate'. 
  Class: traefik
  1. By the name, do you mean the name of the the deployment/pod? If so, that specified with

helm install <name> <helm-folder>

In the docs it is helm install nf .

In this example, nf is the name of the helm deployment, while the deployment is named 'nf-[GUID]'

The GUID is generated to separate different nf-api instances in the name namespace. I suppose the user could just name them differently.

bentsherman commented 4 years ago

Ah I see, all of the hard-coded values I saw were in the values.yaml. Looks good to me then.