cloudberry

This page explains how to enable the Cloudberry kernel in Pigsty for MPP data warehouse scenarios. Cloudberry is integrated through pg_mode: gpsql and the cloudberry package group.

Overview

• Config Name: cloudberry (kernel mode guide, not a standalone conf/*.yml template)
• Node Count: planned by MPP topology (Master/Segment)
• Description: enable Cloudberry kernel with pg_mode: gpsql, while reusing Pigsty deployment, monitoring, and management
• OS Distro: el8, el9, el10, d12, d13, u22, u24
• OS Arch: x86_64, aarch64
• Related: pgsql, mssql, Cloudberry Kernel Guide

How To Enable

In v4.2.0 there is no dedicated ./configure -c cloudberry template yet. Start from meta (or another base template) and switch kernel mode:

./configure -c meta

Then configure pigsty.yml like:

all:
  vars:
    pg_mode: gpsql
    pg_version: 17
    pg_packages: [ cloudberry, pgsql-common ]

Topology Parameters

Cloudberry/Greenplum-style clusters usually require:

• pg_mode: gpsql: enables parallel warehouse mode
• gp_role: master | segment: node role in MPP cluster
• pg_shard: marks the same MPP deployment

For a full topology example, see the gpsql section in pigsty/conf/demo/kernels.yml.

Recommendations

• Prefer PG17 first (current Cloudberry packages target the PG17 ecosystem).
• Complete infrastructure deployment first, then attach Master/Segment nodes in batches.
• Once included in Pigsty monitoring, Cloudberry can use the standard PGSQL dashboards and alerting stack.