Add AFFiNE pod creation script with PostgreSQL and Redis

Shell script to create a rootless Podman pod running AFFiNE v0.26.3
with pgvector/PostgreSQL and Redis, managed via systemd user services.
Exposes AFFiNE Web UI and GraphQL API on port 8092.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

CLAUDE.md

@@ -0,0 +1,12 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Overview
+
+This is a personal `~/bin` utilities repository. It is currently empty and ready for scripts and tools.
+
+## Conventions
+
+- Main branch: `main`
+- Scripts should be executable and include appropriate shebangs

create_pod_affine.sh

@@ -0,0 +1,167 @@
+#!/bin/bash
+
+# To be run by user pln to create pod for AFFiNE with PostgreSQL and Redis
+
+set -e
+
+# Environment variables
+POD_NAME='affine_pod'
+AFFINE_CTR_NAME='affine_ctr'
+AFFINE_MIGRATION_CTR_NAME='affine_migration_ctr'
+POSTGRES_CTR_NAME='affine_postgres_ctr'
+REDIS_CTR_NAME='affine_redis_ctr'
+
+AFFINE_IMAGE='ghcr.io/toeverything/affine:0.26.3'
+POSTGRES_IMAGE='docker.io/pgvector/pgvector:0.7.1-pg16'
+REDIS_IMAGE='docker.io/library/redis:7.4.8'
+
+HOST_LOCAL_IP='127.0.0.1'
+# Expose AFFiNE Web UI on 8092 -> 3010
+AFFINE_HOST_PORT='8092'
+AFFINE_CONTAINER_PORT='3010'
+# All other services are only reachable inside the pod
+
+BIND_DIR="$HOME/.local/share/$POD_NAME"
+POSTGRES_DATA_DIR="$BIND_DIR/postgres-data"
+AFFINE_STORAGE_DIR="$BIND_DIR/affine-storage"
+AFFINE_CONFIG_DIR="$BIND_DIR/affine-config"
+USER_SYSTEMD_DIR="$HOME/.config/systemd/user"
+
+DB_USERNAME='affine'
+DB_PASSWORD='affine'
+DB_DATABASE='affine'
+DATABASE_URL="postgresql://$DB_USERNAME:$DB_PASSWORD@localhost:5432/$DB_DATABASE"
+
+AFFINE_COMMON_ENV=(
+  -e REDIS_SERVER_HOST=localhost
+  -e DATABASE_URL="$DATABASE_URL"
+  -e AFFINE_INDEXER_ENABLED=false
+)
+
+# Stop existing systemd-managed pod if present, to avoid conflicts on rerun
+echo "Stopping systemd-managed pod 'pod-$POD_NAME.service' if it exists..."
+if systemctl --user list-units --type=service --all 2>/dev/null | \
+  grep -q "pod-$POD_NAME.service"; then
+  systemctl --user stop "pod-$POD_NAME.service" || true
+fi
+
+# Prepare directories
+mkdir -p "$POSTGRES_DATA_DIR" "$AFFINE_STORAGE_DIR" "$AFFINE_CONFIG_DIR" \
+  "$USER_SYSTEMD_DIR"
+
+# Create pod if not yet existing
+if ! podman pod exists "$POD_NAME"; then
+  podman pod create -n "$POD_NAME" \
+    -p "$HOST_LOCAL_IP:$AFFINE_HOST_PORT:$AFFINE_CONTAINER_PORT"
+  echo "Pod '$POD_NAME' created (rc=$?)"
+else
+  echo "Pod '$POD_NAME' already exists."
+fi
+
+# Remove any old containers (ignore errors if they don't exist)
+podman rm -f "$AFFINE_CTR_NAME" || true
+podman rm -f "$AFFINE_MIGRATION_CTR_NAME" || true
+podman rm -f "$POSTGRES_CTR_NAME" || true
+podman rm -f "$REDIS_CTR_NAME" || true
+
+# Postgres container (pgvector for AFFiNE)
+podman run -d --name "$POSTGRES_CTR_NAME" --pod "$POD_NAME" \
+  -e POSTGRES_USER="$DB_USERNAME" \
+  -e POSTGRES_PASSWORD="$DB_PASSWORD" \
+  -e POSTGRES_DB="$DB_DATABASE" \
+  -e POSTGRES_INITDB_ARGS='--data-checksums' \
+  -v "$POSTGRES_DATA_DIR:/var/lib/postgresql/data:Z" \
+  "$POSTGRES_IMAGE"
+echo "Container '$POSTGRES_CTR_NAME' started (rc=$?)"
+
+# Wait for Postgres to be ready
+echo "Waiting for Postgres to be ready (pg_isready)..."
+for attempt in $(seq 1 30); do
+  if podman exec "$POSTGRES_CTR_NAME" pg_isready -U "$DB_USERNAME" -d "$DB_DATABASE" \
+    >/dev/null 2>&1; then
+    echo "Postgres is ready."
+    break
+  fi
+  sleep 2
+  if [ "$attempt" -eq 30 ]; then
+    echo "ERROR: Postgres did not become ready in time." >&2
+    exit 1
+  fi
+done
+
+# Redis container
+podman run -d --name "$REDIS_CTR_NAME" --pod "$POD_NAME" \
+  "$REDIS_IMAGE"
+echo "Container '$REDIS_CTR_NAME' started (rc=$?)"
+
+# Wait for Redis to be ready
+echo "Waiting for Redis to be ready..."
+for attempt in $(seq 1 30); do
+  if podman exec "$REDIS_CTR_NAME" redis-cli ping >/dev/null 2>&1; then
+    echo "Redis is ready."
+    break
+  fi
+  sleep 2
+  if [ "$attempt" -eq 30 ]; then
+    echo "ERROR: Redis did not become ready in time." >&2
+    exit 1
+  fi
+done
+
+# AFFiNE migration job (must complete before starting the server)
+echo "Running AFFiNE database migration..."
+podman run --rm --name "$AFFINE_MIGRATION_CTR_NAME" --pod "$POD_NAME" \
+  "${AFFINE_COMMON_ENV[@]}" \
+  -v "$AFFINE_STORAGE_DIR:/root/.affine/storage:Z" \
+  -v "$AFFINE_CONFIG_DIR:/root/.affine/config:Z" \
+  "$AFFINE_IMAGE" \
+  sh -c 'node ./scripts/self-host-predeploy.js'
+echo "AFFiNE migration completed (rc=$?)"
+
+# AFFiNE server container
+podman run -d --name "$AFFINE_CTR_NAME" --pod "$POD_NAME" \
+  "${AFFINE_COMMON_ENV[@]}" \
+  -v "$AFFINE_STORAGE_DIR:/root/.affine/storage:Z" \
+  -v "$AFFINE_CONFIG_DIR:/root/.affine/config:Z" \
+  "$AFFINE_IMAGE"
+echo "Container '$AFFINE_CTR_NAME' started (rc=$?)"
+
+# Generate systemd service files
+cd "$USER_SYSTEMD_DIR"
+podman generate systemd --name --new --files "$POD_NAME"
+echo "Generated systemd service files (rc=$?)"
+
+# Stop & remove live pod and containers
+podman pod stop --ignore --time 15 "$POD_NAME"
+podman pod rm -f --ignore "$POD_NAME"
+if podman pod exists "$POD_NAME"; then
+  echo "ERROR: Pod $POD_NAME still exists." >&2
+  exit 1
+else
+  echo "Stopped & removed live pod $POD_NAME and containers."
+fi
+
+# Enable systemd user services
+systemctl --user daemon-reload
+# pod service (creates pod + containers)
+systemctl --user enable --now "pod-${POD_NAME}.service"
+systemctl --user is-enabled "pod-${POD_NAME}.service"
+systemctl --user is-active "pod-${POD_NAME}.service"
+echo "Enabled systemd service pod-${POD_NAME}.service (rc=$?)"
+echo "To view status: systemctl --user status pod-${POD_NAME}.service"
+echo "To view logs: journalctl --user -u pod-${POD_NAME}.service -f"
+
+# Wait for AFFiNE Web UI readiness
+CHECK_URL="http://$HOST_LOCAL_IP:$AFFINE_HOST_PORT"
+for attempt in $(seq 1 30); do
+  if curl -fsS "$CHECK_URL" >/dev/null 2>&1; then
+    echo "AFFiNE Web UI is reachable at $CHECK_URL."
+    echo "AFFiNE GraphQL API is reachable at $CHECK_URL/graphql."
+    break
+  fi
+  sleep 2
+  if [ "$attempt" -eq 30 ]; then
+    echo "timeout error." >&2
+    exit 1
+  fi
+done

prompt_shell-script.md

@@ -0,0 +1,64 @@
+# Podman shell script for AFFiNE service
+
+## Motivation
+
+A Podman Pod is needed on a Linux server which provides an AFFiNE service.
+This Podman Pod needs to be created and started with a shell script which needs to be designed.
+
+The AFFiNE service will be used by humans and by AI agents; it will also be part of a design environment for designing a special MCP server for AFFiNE.
+
+## Shell script requirements
+
+Requirements for the shell script:
+- Must use a container image with a Pinned Tag (for exact consistency) which points to the AFFiNE version 0.26.3. The name likely is something like "ghcr.io/toeverything/affine:0.26.3".
+- Must provide the AFFiNE web user interface at port 8092.
+- Must provide the GraphQL API. Background: The AFFiNE web and desktop apps use an internal GraphQL API to communicate with the backend. There is a /graphql endpoint but it is not documented for third-party use.
+- Must be in folder /home/pln/bin.
+- Must have the name create_pod_affine.sh
+
+The shell script shall be run by user pln which has permissions to run rootless pods.
+
+## Shell script style
+
+The needed shell script must have the same style as other shell scripts on the server.
+
+These files are examples:
+/home/lwc/bin/create_pod_langflow.sh
+/home/krt/bin/create_pod_qdrant.sh
+
+## Your tasks
+
+### Ask first
+
+Before starting to design the shell script, ask between two and five questions to fully understand the situation, your tasks and the objectives.
+
+### Identify the container image
+
+Find the container image with Pinned Tag pointing to AFFiNE version 0.26.3.
+
+### Write the shell script
+
+Write the shell script.
+
+### Test the shell script
+
+Run the shell script and test it.
+
+### Redesign if necessary
+
+If the test failed, understand the problem, improve the shell script and go back to Test the shell script.
+
+Repeat this in a loop up to five times.
+
+## Your objectives
+
+Your objectives are:
+- All requirements are fulfilled.
+- AFFiNE web user interface shows up at 127.0.0.1:8092.
+- The AFFiNE GraphQL API shows up under 127.0.0.1:8092 at /graphql or another link.
+
+## Your behaviour
+
+If it is not possible to achieve your objectives, interrupt and ask me.
+
+Complete all your tasks without asking in between if you can achieve your objectives.