From 560f291db97ceb19bf9e5f6190ef737a26954f42 Mon Sep 17 00:00:00 2001
From: Florent Guiotte <florent.guiotte@irisa.fr>
Date: Sun, 11 Aug 2024 21:59:44 +0200
Subject: [PATCH] Add backup script

---
 README.md |   8 ++++
 backup.py | 110 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 118 insertions(+)
 create mode 100755 backup.py

diff --git a/README.md b/README.md
index 7c13b86..467d5ea 100644
--- a/README.md
+++ b/README.md
@@ -45,6 +45,14 @@ dcc restart nextcloud-db
 dcc up -d nextcloud
 ```
 
+## Backup
+
+To backup the named volumes, run:
+
+```shell
+./backup.py
+```
+
 ## Logs
 
     2022-11-02 Created OVH token
diff --git a/backup.py b/backup.py
new file mode 100755
index 0000000..c0b06f6
--- /dev/null
+++ b/backup.py
@@ -0,0 +1,110 @@
+#!/usr/bin/env python
+# file backup.py
+# author Florent Guiotte <florent.guiotte@irisa.fr>
+# version 0.0
+# date 10 août 2024
+"""Abstract
+
+doc.
+"""
+
+import yaml
+from pathlib import Path
+import subprocess
+from datetime import datetime
+
+
+COMPOSE_PATH = Path('docker-compose.yml')
+BACKUP_PATH = Path('/mnt/storage/docker-bkp')
+VOLUME_PREFIX = 'docker_'
+
+
+class UnionFind:
+    def __init__(self):
+        self.parent = {}
+
+    def make(self, service):
+        self.parent.setdefault(service, service)
+
+    def find(self, service):
+        """return root"""
+        if self.parent[service] != service:
+            self.parent[service] = self.find(self.parent[service])
+
+        return self.parent[service]
+
+    def union(self, service1, service2):
+        root1 = self.find(service1)
+        root2 = self.find(service2)
+        if root1 != root2:
+            self.parent[root2] = root1 # compress!
+
+def build_services_graph(services):
+    uf = UnionFind()
+
+    for service in services:
+        uf.make(service)
+        for dependency in services[service].get('depends_on', []):
+            uf.make(dependency)
+            uf.union(service, dependency)
+
+    return uf
+
+def group_services(services, graph):
+    grouped_services = {}
+    for service in services:
+        root = graph.find(service)
+        if root not in grouped_services:
+            grouped_services[root] = {'services': []}
+        grouped_services[root]['services'].append(service)
+
+    return grouped_services
+
+
+def group_volumes(services, volumes, services_group):
+    for group_name, group in services_group.items():
+        group_volumes = group.setdefault('volumes', [])
+        for service in group['services']:
+            for volume in [v.split(':')[0] for v in services[service]['volumes']]:
+                if volume in volumes: group_volumes += [volume]
+
+    return services_group
+
+def backup(volume):
+    current_date = datetime.now()
+    date_string = current_date.strftime("%Y-%m-%d")
+    archive_name = f'{date_string}_{volume}.tar'
+    print(f'backup volume {volume} to {BACKUP_PATH}/{archive_name}')
+
+    subprocess.run(f'docker run --rm --volume {VOLUME_PREFIX}{volume}:/data --volume {BACKUP_PATH}:/bkp ubuntu tar -cf /bkp/{archive_name} -C /data .'.split())
+
+
+def run_docker_compose(cmd):
+    subprocess.run(f'docker compose {cmd}'.split())
+
+
+if __name__ == '__main__':
+    with COMPOSE_PATH.open() as cf:
+        compose = yaml.safe_load(cf)
+
+    services = compose['services']
+    volumes = compose['volumes']
+
+
+    services_graph = build_services_graph(services)
+    services_group = group_services(services, services_graph)
+    services_group = group_volumes(services, volumes, services_group)
+
+    for group_name, group in services_group.items():
+        print(f'Service group {group_name} ', end='')
+
+        if not group['volumes']:
+            print('no volumes')
+            continue
+
+        print('run backup...')
+
+        run_docker_compose(f'stop {" ".join(group["services"])}')
+        for volume in group['volumes']:
+            backup(volume)
+        run_docker_compose(f'start {" ".join(group["services"])}')