Running Multiple Pulp Container Instances Behind a Single Endpoint for Load Handling

Is it officially supported to run multiple Pulp Container instances behind a single endpoint to handle higher pull/load traffic?

If yes:

  • Are there any specific requirements or constraints to support this?
  • Are there any recommended/supported reference architectures for this setup?
  • Any documentation available for the same

If not:

  • Is it on the Pulp roadmap and will be available as a feature in future?

Yes, it is possible by design.
Documentation is here:
https://pulpproject.org/pulpcore/docs/admin/learn/architecture/

Important are the things that need to be shared:

  • Storage unless you divert to a blockstorage technology.
  • Configuration
  • Database
  • [optional] redis
1 Like

I do this with 3 totally independent instances, nothing shared between them and a floating VIP for the public access. The instances are kept in perfect sync via other mechanisms

3 Likes