Skip to content

Central sets WORKER_COUNT based on host, not container resources #577

Closed
@yanokwa

Description

In https://github.com/getodk/central/blob/HEAD/files/service/scripts/start-odk.sh#L33 we set WORKER_COUNT based on total memory.

According to https://forum.getodk.org/t/limiting-resources-memory-that-the-odk-central-service-can-consume/44357/2 that total memory is the host's memory not the container's memory, so you can have a situation where you are running Central in a memory limited container but using more workers than is recommended.

My guess is that most people don't allocate less than 1 GB, so this isn't urgent, but I wanted to document it.

Metadata

Assignees

No one assigned

    Labels

    help wantedopsDocker, nginx, ops to deploy Central

    Type

    No type

    Projects

    • Status

      ✅ done

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions