mirror of https://github.com/hanzoai/postgres.git synced 2026-10-30 18:33:07 +00:00

Files

T

zooqueenandHanzo Dev bbdb14f0bf docs: hanzo/iam is not a fork of anything, and its config is not that one

hanzoai/iam is original clean-room work — its LICENSE states it carries no
third-party licensed source. The derived identity server was hanzoai/iam-v1,
and it is retired. Calling the IAM we ship a vendor fork was factually wrong,
not just badly worded.

The connection example under that heading was the retired server's conf
format (driverName / dataSourceName), which current IAM does not read. It
now shows what the code actually does: internal/store/open.go selects the
backend from the --store flag (sqlite | sql | datastore) and the sql backend
reaches this server over ZAP, not a libpq DSN.

Co-authored-by: Hanzo Dev <[email protected]>

2026-07-28 17:14:30 -07:00

2.3 KiB

Raw Permalink Blame History

Hanzo Relational (PostgreSQL Fork)

Overview

Hanzo Relational is a fork of PostgreSQL optimized for the Hanzo AI platform's transactional workloads. It provides:

Primary OLTP Database - Core business data
pgvector Extension - Vector embeddings for AI
Multi-tenant Support - Organization-scoped schemas
High Availability - Streaming replication ready

Repository: https://github.com/hanzoai/sql

Quick Start

# Start PostgreSQL with Hanzo config
cd hanzo
docker compose up -d

# Connect to PostgreSQL
docker exec -it hanzo-postgres psql -U hanzo -d hanzo

Hanzo Extensions

Pre-configured extensions:

pgvector - Vector similarity search
pg_stat_statements - Query performance
uuid-ossp - UUID generation
pg_trgm - Trigram text search
btree_gin - GIN index support

Integration Points

With hanzo/console (LangFuse fork)

Console uses PostgreSQL for metadata:

DATABASE_URL=postgresql://hanzo:password@localhost:5432/hanzo

With hanzo/iam

IAM stores user/org data. The store backend is chosen by flag, and sql reaches this server over ZAP (:9651), not a libpq DSN:

iam --store sql   # sqlite | sql | datastore

With hanzo/base (PocketBase fork)

Base instances use embedded SQLite, but orchestration metadata in PostgreSQL.

Syncing with Upstream

# Fetch upstream changes
git fetch upstream

# Merge upstream master
git merge upstream/master

# Keep hanzo/ directory
git checkout --ours hanzo/

git push origin master

Performance Tuning

Recommended Settings (16GB RAM)

-- postgresql.conf
shared_buffers = 4GB
effective_cache_size = 12GB
maintenance_work_mem = 1GB
checkpoint_completion_target = 0.9
wal_buffers = 64MB
default_statistics_target = 100
random_page_cost = 1.1
effective_io_concurrency = 200
work_mem = 52428kB
min_wal_size = 1GB
max_wal_size = 4GB
max_worker_processes = 8
max_parallel_workers_per_gather = 4
max_parallel_workers = 8
max_parallel_maintenance_workers = 4

Docker Compose

See hanzo/compose.yml for local development with:

PostgreSQL 17
pgAdmin for management
pgvector pre-installed

hanzo/console - AI observability (uses for metadata)
hanzo/iam - Identity management (users, orgs)
hanzo/datastore - ClickHouse fork (OLAP)
hanzo/memory - Redis fork (caching)

2.3 KiB Raw Permalink Blame History