Harbor: Kubernetes Operator for Harbor

Created on 12 Jun 2019  路  5Comments  路  Source: goharbor/harbor

Create an operator for Harbor

requirements TBD - influenced by a POC/Design
Some examples
-Setup and operations of Harbor
-Backup and recovery
-Scaling
-Setup helm chart components like postgres and redis

arek8s-support areoperator kinrequirement

Most helpful comment

Hello @michmike,

At OVH we currently have an operator. We would like to open source it. @xaleeks and us started to work on it.

So you know, here is the current state of the operator:

  • Setup Harbor (thanks to a CustomResourceDefinition)
  • Deploy independently Clair, ChartMuseum.
  • Components configurations (such as registry backend) is configured with Kubernetes secret.
  • We removed the static nginx proxy pass configured on the Harbor Helm chart to use native ingress.
  • Databases are provisioned thanks to external resource: operator will look for secret containing database credentials (or redis credentials).
    We considered databases should be handled by their own operator and/or managed services, those exist ou should exist for most common DBMS.
    These operator would also be in charge of backups.
  • About scaling, currently we designed it to create deployments where replicas number is customizable, do you expect something more?

Please feedback is appreciated :)

All 5 comments

Hello @michmike,

At OVH we currently have an operator. We would like to open source it. @xaleeks and us started to work on it.

So you know, here is the current state of the operator:

  • Setup Harbor (thanks to a CustomResourceDefinition)
  • Deploy independently Clair, ChartMuseum.
  • Components configurations (such as registry backend) is configured with Kubernetes secret.
  • We removed the static nginx proxy pass configured on the Harbor Helm chart to use native ingress.
  • Databases are provisioned thanks to external resource: operator will look for secret containing database credentials (or redis credentials).
    We considered databases should be handled by their own operator and/or managed services, those exist ou should exist for most common DBMS.
    These operator would also be in charge of backups.
  • About scaling, currently we designed it to create deployments where replicas number is customizable, do you expect something more?

Please feedback is appreciated :)

thank you for all the work your team did on this @holyhope

@holyhope @xaleeks Do you have the release timeline? I'm willing to test it even in the development stage.

@lewcpe this is not in scope for Harbor 2.0 (ETA april 2020). it will likely make it to the next release of Harbor. if you have time and resources to help us develop the operator, that would help a lot. please sync with @xaleeks

Was this page helpful?
0 / 5 - 0 ratings

Related issues

protochron picture protochron  路  3Comments

mramanathan picture mramanathan  路  3Comments

abououdine picture abououdine  路  3Comments

reasonerjt picture reasonerjt  路  3Comments

a-kinder picture a-kinder  路  3Comments