Skip to content

BACKUP_RESTORE_GUIDE.md

Latest commit

 

History

History
176 lines (121 loc) · 6.37 KB

BACKUP_RESTORE_GUIDE.md

File metadata and controls

176 lines (121 loc) · 6.37 KB
Raw

Backup and Restore Guide

This document describes the backup and restore procedures for the Med_SIS GeoNode/GeoServer stack. All scripts are Bash-based and use Docker Compose.

Overview

The system performs a full backup of GeoNode metadata, GeoServer configuration, and PostgreSQL databases. Restore can target either:

  • Test environment (s4m_catalogue_test) — isolated from production, uses port 8081
  • Production environment (s4m_catalogue) — live system on port 80

Files Involved

File Role
lib_common.sh Shared functions (DB wait, restore, GeoServer fixes, etc.)
complete_backup.sh Full backup: GeoNode + GeoServer + databases
restore_on_test.sh Restore to isolated test environment
restore_on_production.sh Restore to production environment
cleanup_test.sh Stops test stack, removes artifacts, restarts production

Backup Procedure

Script: complete_backup.sh

Usage:

./complete_backup.sh

Prerequisites: Production stack must be running (docker-compose up -d).

What it does:

  1. Preventive restart — Restarts Django and GeoServer for a clean state.
  2. GeoNode backup — Runs manage.py backup (official GeoNode backup). Output: YYYYMMDD-HHMMSS.zip in ./backups/.
  3. GeoServer datastores — Archives all datastore.xml files from GeoServer workspaces. Output: geoserver-datastores-YYYYMMDD-HHMMSS.tar.gz.
  4. Geodata dump — pg_dump of my_geo_data (shapefile tables). Output: backup-geodata-YYYYMMDD-HHMMSS.dump.
  5. Backoffice dump — pg_dump of backoffice. Output: backup-backoffice-YYYYMMDD-HHMMSS.dump.

Output directory: ./backups/

Restore Procedure

Test Environment: restore_on_test.sh

Usage:

./restore_on_test.sh                                    # Auto-discover latest backups in ./backups/
./restore_on_test.sh /path/to/backup.zip /path/to/backoffice.dump   # Explicit files

Behaviour:

  • Auto-discovery: With no arguments, searches ./backups/ (or $BACKUP_SEARCH_DIR) for the most recent .zip, backup-backoffice-*.dump, backup-geodata-*.dump, and geoserver-datastores-*.tar.gz.
  • Explicit files: With two arguments, uses the given GeoNode backup and backoffice dump; geodata and datastores are auto-discovered in the same directory as the GeoNode backup.

Phases:

  1. Shutdown — Stops production containers on conflicting ports (nginx, geoserver), then stops the test stack.
  2. Data restore — Prepares test DBs (test_my_geo, test_my_geo_data, test_backoffice), restores backoffice and geodata dumps, extracts GeoServer catalog from the GeoNode zip, runs migrations, injects datastores (with DB name substitution), runs manage.py restore, fixes URLs in the DB, and runs migrate_baseurl.
  3. Startup — Starts the full test stack, waits for Django health, and applies GeoServer styles fix.
  4. Verification — Prints status and URLs.

Test URL: http://localhost:8081
Production: Unchanged at http://localhost

Production Environment: restore_on_production.sh

Usage:

./restore_on_production.sh                              # Auto-discover latest backups
./restore_on_production.sh /path/to/backup.zip /path/to/backoffice.dump   # Explicit files

Behaviour:

  • Same auto-discovery and explicit-file logic as the test restore.

Phases:

  1. Partial shutdown — Stops Django, GeoNode, Celery, GeoServer, data-dir-conf. Database stays up.
  2. DB preparation — Prepares my_geo, my_geo_data, backoffice.
  3. Database restore — Restores backoffice and geodata dumps.
  4. Migration + GeoNode restore — Runs migrations, starts GeoServer, runs manage.py restore, injects datastores (no DB name substitution).
  5. Full restart — Starts all services and applies GeoServer styles fix.

Cleanup: cleanup_test.sh

Usage:

./cleanup_test.sh

What it does:

  1. Stops the test environment and removes volumes.
  2. Deletes ./geoserver_data_dir_test.
  3. Renames docker-compose.override.yml → docker-compose.override.yml.backup (so production does not use test override).
  4. Restarts the production stack.

Shared Library: lib_common.sh

Usage: All scripts source it with:

source "$(dirname "$0")/lib_common.sh"

Main functions:

Function Purpose
find_latest_file DIR PATTERN Returns the most recent file matching the pattern in DIR
wait_for_db CONTAINER [USER] [MAX] Waits for PostgreSQL to accept connections
grant_db_permissions CONTAINER DB OWNER [USER] Grants schema/table permissions to the application user
fix_fixture_fk_violations CONTAINER DB [USER] Removes orphan records that block fixture loading
prepare_db CONTAINER DB OWNER [USER] Drops, recreates, enables PostGIS, sets permissions
restore_pg_dump CONTAINER DUMP DB [USER] Copies dump into container and runs pg_restore
apply_manage_restore COMPOSE DJANGO CONTAINER ZIP CONFIG Runs manage.py restore with a curs.close() bug patch
wait_for_geoserver CONTAINER [MAX] Waits for GeoServer to respond (HTTP 200/302/401/403)
fix_geoserver_styles_after_startup CONTAINER Fixes corrupted GeoServer styles (backup SLD, reload catalog, register via REST, fix layer associations)
inject_geoserver_datastores CONTAINER ARCHIVE [DB_MAP] Injects datastore.xml files; optionally substitutes DB names for test

Backup File Naming

Pattern Purpose
*.zip GeoNode backup (metadata, layers, GeoServer catalog)
backup-backoffice-*.dump Backoffice PostgreSQL dump
backup-geodata-*.dump my_geo_data (shapefile tables) dump
geoserver-datastores-*.tar.gz GeoServer datastore.xml backup

Environment Variables

Variable Default Used by
BACKUP_SEARCH_DIR ./backups restore_on_test.sh (auto-discovery)
SITEURL http://localhost restore_on_production.sh (final message)
COMPOSE_PROJECT_NAME s4m_catalogue / s4m_catalogue_test All scripts

Quick Reference

# Backup (production must be running)
./complete_backup.sh

# Restore to test
./restore_on_test.sh

# Restore to production
./restore_on_production.sh

# Cleanup test and restart production
./cleanup_test.sh