install_lakekeeper

The install_lakekeeper role installs Lakekeeper, the Iceberg REST catalog that ColdFront's cold tier commits through, along with the Postgres packages for the catalog's own instance, which setup_lakekeeper creates. Lakekeeper is not a pgEdge Distributed Postgres node: its Postgres instance exists for the catalog alone and never runs Spock.

The role performs the following tasks on inventory hosts:

• Install the pgedge-lakekeeper package, the Postgres packages for pg_version, and pgedge-python3-psycopg2.

Role Dependencies

This role requires the following roles for normal operation:

• role_config provides shared configuration variables to the role.
• init_server prepares the host and checks the inventory, including that the host's ports do not collide.
• install_repos configures pgEdge package repositories.

When to Use

Execute this role on a dedicated lakekeeper host, separate from any pgEdge Distributed Postgres node. Lakekeeper's own architecture notes call for the catalog to live on its own instance, "never co-located on a ColdFront data node," since co-locating couples the catalog's availability and load to a data node.

- hosts: lakekeeper
  collections:
    - pgedge.platform
  roles:
    - install_repos
    - install_lakekeeper
    - setup_lakekeeper

Apply init_server to the Lakekeeper host first, as to every host. The sample-playbooks/simple-cluster-coldfront playbook does so in the same play, and sample-playbooks/ultra-ha-coldfront in its first play, which runs it on every host:

- hosts: lakekeeper
  collections:
    - pgedge.platform
  roles:
    - init_server
    - install_repos
    - install_lakekeeper
    - setup_lakekeeper

Configuration

This role uses the following parameters from the inventory file:

The settings the catalog itself runs on, such as its ports, its database and its Iceberg namespace, belong to setup_lakekeeper.

See the Configuration Reference for descriptions and defaults.

How It Works

1. Installs pgedge-lakekeeper and the Postgres package matching pg_version. Package names differ by OS family: Debian nests the version between postgresql and the package name (pgedge-postgresql-<ver>); RHEL suffixes the version onto the base name instead, and splits client tools (pgedge-postgresql<ver>) from the server binaries (pgedge-postgresql<ver>-server), both of which this role installs. On RHEL it also installs the contrib package, which holds the extensions Lakekeeper's migration needs.

Artifacts

Platform-Specific Behavior

Confirmed against the packaged file lists on apt.pgedge.com and dnf.pgedge.com: on RHEL, the base pgedge-postgresql<ver> package is client tools only (psql, pg_dump, and similar); the server binaries (postgres, initdb, pg_ctl) live in the separate pgedge-postgresql<ver>-server package, which this role also installs. Debian's pgedge-postgresql-<ver> package is the server itself, with no such split.

Idempotency

This role is idempotent and safe to re-run on inventory hosts. Package installation is a no-op once already done.