Backup & Restore¶

Automated Postgres dumps + config file backups, uploaded to S3 (or any rclone-supported remote like Google Drive).

Prerequisites¶

Docker & Docker Compose (already required by the project)
rclone — sudo pacman -S rclone (Arch) or curl https://rclone.org/install.sh | sudo bash

Setup¶

1. Configure S3 credentials¶

No rclone config file needed — the scripts pass credentials via environment variables. Add these to your .env or export them in your shell:

AWS_ACCESS_KEY_ID=your-key
AWS_SECRET_ACCESS_KEY=your-secret
AWS_REGION=us-east-1
RCLONE_REMOTE=:s3:your-bucket-name

Note: The :s3: prefix is rclone's connection string syntax — it tells rclone to use the S3 backend without a named config entry.

Create the bucket if it doesn't exist:

aws s3 mb s3://your-bucket-name --region us-east-1

Verify it works:

export AWS_ACCESS_KEY_ID=your-key AWS_SECRET_ACCESS_KEY=your-secret RCLONE_REMOTE=:s3:your-bucket-name
rclone lsd "$RCLONE_REMOTE"

2. Run a manual backup¶

./scripts/backup.sh

This will: 1. pg_dump the database via docker compose exec 2. Bundle the dump with .env, bots/, skills/, mcp.yaml, and config/searxng/settings.yml 3. Upload the tarball to the rclone remote (RCLONE_REMOTE) 4. Prune local backups to the most recent 7

3. Set up cron (daily at 2 AM)¶

crontab -e

Add:

0 2 * * * /path/to/scripts/backup.sh >> /var/log/spindrel-backup.log 2>&1

Restore¶

From the latest remote backup¶

./scripts/restore.sh

This pulls the latest archive from the rclone remote, extracts config files, restores the database, and restarts the stack.

From a specific local archive¶

./scripts/restore.sh ./backups/agent-backup-20260322_030000.tar.gz

Database-only restore¶

If config files haven't changed and you only need to roll back data:

docker compose exec -T postgres pg_restore -U agent -d agentdb \
  --clean --if-exists --no-owner < ./backups/agentdb_YYYYMMDD_HHMMSS.dump
docker compose restart agent-server

Configuration¶

Both scripts support environment variable overrides:

Variable	Default	Description
`AWS_ACCESS_KEY_ID`	(required)	AWS access key for S3
`AWS_SECRET_ACCESS_KEY`	(required)	AWS secret key for S3
`AWS_REGION`	`us-east-1`	AWS region
`RCLONE_REMOTE`	(required)	rclone remote path (e.g. `:s3:your-bucket-name`)
`BACKUP_DIR`	`./backups`	Local directory for backup archives
`LOCAL_KEEP`	`7`	Number of local backups to retain (backup.sh only)

What's backed up¶

Asset	Why
Postgres database (`pg_dump -Fc`)	All agent state: messages, memories, knowledge, sessions, etc.
`.env`	Runtime config and API keys
`bots/*.yaml`	Bot configurations
`skills/*.md`	Skill definitions
`mcp.yaml`	MCP server config
`config/searxng/settings.yml`	SearXNG customization

Migration to a new server¶

See docs/migration.md for the full migration guide, including deployment mode switching (native ↔ Docker), Cloudflare tunnel setup, and integration re-pointing.

Quick version:

On old host: ./scripts/backup.sh
On new host: clone repo, install Docker + rclone, set the same AWS/rclone env vars
Run ./scripts/restore.sh
Review .env for host-specific values (LITELLM_BASE_URL, paths, etc.)
Set up cron on the new host
Re-point Cloudflare tunnel connector (install cloudflared on new host)