Support for rsync+ssh backup target (#678)

* Added support for backup to a remote server using rsync

* updated web interface to get data from user
* added way to list files from server

It’s not using the “username” field of the yaml configuration
file to minimise the amount of patches needed. So the username
is actually sorted within the rsync URL.

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* Added ssh key generation upon installation for root user.

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* Removed stale blank lines, and fixed typo

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* fix backup-location lines, by switching it from id to class

* Various web UI fixes

- fixed user field being shadowed ;
- fixed settings reading comparaison ;
- fixed forgotten min-age field.

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* Added SSH Public Key shown on the web interface UI

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* trailing spaces.

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* fixed the extraneous environment

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* Updated key setup

- made key lower in bits, but stronger (using -a option),
- made ssh-keygen run in background using nohup,
- added independent key file, as id_rsa_miab,
- added ssh-options to all duplicity calls to use the id_rsa_miab keyfile,
- changed path to the public key display

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* added rsync options for ssh identity support

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* removed strict host checking for all backup operations

Signed-off-by: Bernard `Guyzmo` Pratz <guyzmo+github@m0g.net>

* Remove nohup from ssh-keygen so errors aren't hidden. Also only generate a key if none exists yet

* Add trailing slash when checking a remote backup. Also check if we actually can read the remote size

* Factorisation of the repeated rsync/ssh options

cf https://github.com/mail-in-a-box/mailinabox/pull/678#discussion_r81478919

* Updated message SSH key creation

https://github.com/mail-in-a-box/mailinabox/pull/678#discussion_r81478886

management/backup.py

@@ -30,6 +30,11 @@ def backup_status(env):
 	backups = { }
 	backup_cache_dir = os.path.join(backup_root, 'cache')
 
+	rsync_ssh_options = [
+		"--ssh-options='-i /root/.ssh/id_rsa_miab'",
+		"--rsync-options=-e \"/usr/bin/ssh -oStrictHostKeyChecking=no -oBatchMode=yes -p 22 -i /root/.ssh/id_rsa_miab\"",
+	]
+
 	def reldate(date, ref, clip):
 		if ref < date: return clip
 		rd = dateutil.relativedelta.relativedelta(ref, date)
@@ -52,6 +57,7 @@ def backup_status(env):
 			"size": 0, # collection-status doesn't give us the size
 			"volumes": keys[2], # number of archive volumes for this backup (not really helpful)
 		}
+
 	code, collection_status = shell('check_output', [
 		"/usr/bin/duplicity",
 		"collection-status",
@@ -59,7 +65,7 @@ def backup_status(env):
 		"--gpg-options", "--cipher-algo=AES256",
 		"--log-fd", "1",
 		config["target"],
-		],
+		] + rsync_ssh_options,
 		get_env(env),
 		trap=True)
 	if code != 0:
@@ -204,7 +210,7 @@ def perform_backup(full_backup):
 	backup_cache_dir = os.path.join(backup_root, 'cache')
 	backup_dir = os.path.join(backup_root, 'encrypted')
 
-	# Are backups dissbled?
+	# Are backups disabled?
 	if config["target"] == "off":
 		return
 
@@ -283,7 +289,7 @@ def perform_backup(full_backup):
 			env["STORAGE_ROOT"],
 			config["target"],
 			"--allow-source-mismatch"
-			],
+			] + rsync_ssh_options,
 			get_env(env))
 	finally:
 		# Start services again.
@@ -305,7 +311,7 @@ def perform_backup(full_backup):
 		"--archive-dir", backup_cache_dir,
 		"--force",
 		config["target"]
-		],
+		] + rsync_ssh_options,
 		get_env(env))
 
 	# From duplicity's manual:
@@ -320,7 +326,7 @@ def perform_backup(full_backup):
 		"--archive-dir", backup_cache_dir,
 		"--force",
 		config["target"]
-		],
+		] + rsync_ssh_options,
 		get_env(env))
 
 	# Change ownership of backups to the user-data user, so that the after-bcakup
@@ -359,7 +365,7 @@ def run_duplicity_verification():
 		"--exclude", backup_root,
 		config["target"],
 		env["STORAGE_ROOT"],
-	], get_env(env))
+	] + rsync_ssh_options, get_env(env))
 
 def run_duplicity_restore(args):
 	env = load_environment()
@@ -370,7 +376,7 @@ def run_duplicity_restore(args):
 		"restore",
 		"--archive-dir", backup_cache_dir,
 		config["target"],
-		] + args,
+		] + rsync_ssh_options + args,
 	get_env(env))
 
 def list_target_files(config):
@@ -383,6 +389,34 @@ def list_target_files(config):
 	if p.scheme == "file":
 		return [(fn, os.path.getsize(os.path.join(p.path, fn))) for fn in os.listdir(p.path)]
 
+	elif p.scheme == "rsync":
+		rsync_fn_size_re = re.compile(r'.*    ([^ ]*) [^ ]* [^ ]* (.*)')
+		rsync_target = '{host}:{path}'
+
+		_, target_host, target_path = config['target'].split('//')
+		target_path = '/' + target_path
+		if not target_path.endswith('/'):
+			target_path += '/'
+
+		rsync_command = [ 'rsync',
+					'-e',
+					'/usr/bin/ssh -i /root/.ssh/id_rsa_miab -oStrictHostKeyChecking=no -oBatchMode=yes',
+					'--list-only',
+					'-r',
+					rsync_target.format(
+						host=target_host,
+						path=target_path)
+				]
+
+		code, listing = shell('check_output', rsync_command, trap=True)
+		if code == 0:
+			for l in listing.split('\n'):
+				match = rsync_fn_size_re.match(l)
+				if match:
+					yield (match.groups()[1], int(match.groups()[0].replace(',','')))
+		else:
+			raise ValueError("Connection to rsync host failed")
+
 	elif p.scheme == "s3":
 		# match to a Region
 		fix_boto() # must call prior to importing boto
@@ -482,6 +516,9 @@ def get_backup_config(env, for_save=False, for_ui=False):
 	if config["target"] == "local":
 		# Expand to the full URL.
 		config["target"] = "file://" + config["file_target_directory"]
+	ssh_pub_key = os.path.join('/root', '.ssh', 'id_rsa_miab.pub')
+	if os.path.exists(ssh_pub_key):
+		config["ssh_pub_key"] = open(ssh_pub_key, 'r').read()
 
 	return config
 

management/templates/system-backup.html

@@ -16,16 +16,60 @@
       <select class="form-control" rows="1" id="backup-target-type" onchange="toggle_form()">
         <option value="off">Nowhere (Disable Backups)</option>
         <option value="local">{{hostname}}</option>
+        <option value="rsync">rsync</option>
         <option value="s3">Amazon S3</option>
       </select>
     </div>
   </div>
+  <!-- LOCAL BACKUP -->
   <div class="form-group backup-target-local">
     <div class="col-sm-10 col-sm-offset-2">
-      <p>Backups are stored on this machine&rsquo;s own hard disk. You are responsible for periodically using SFTP (FTP over SSH) to copy the backup files from <tt id="backup-location"></tt> to a safe location. These files are encrypted, so they are safe to store anywhere.</p>
+      <p>Backups are stored on this machine&rsquo;s own hard disk. You are responsible for periodically using SFTP (FTP over SSH) to copy the backup files from <tt class="backup-location"></tt> to a safe location. These files are encrypted, so they are safe to store anywhere.</p>
       <p>Separately copy the encryption password from <tt class="backup-encpassword-file"></tt> to a safe and secure location. You will need this file to decrypt backup files.</p>
     </div>
   </div>
+  <!-- RSYNC BACKUP -->
+  <div class="form-group backup-target-rsync">
+    <div class="col-sm-10 col-sm-offset-2">
+
+      <p>Backups synced to a remote machine using rsync over SSH, with local
+      copies in <tt class="backup-location"></tt>. These files are encrypted, so
+      they are safe to store anywhere.</p> <p>Separately copy the encryption
+      password from <tt class="backup-encpassword-file"></tt> to a safe and
+      secure location. You will need this file to decrypt backup files.</p>
+
+    </div>
+  </div>
+  <div class="form-group backup-target-rsync">
+    <label for="backup-target-rsync-host" class="col-sm-2 control-label">Hostname</label>
+    <div class="col-sm-8">
+      <input type="text" placeholder="hostname.local" class="form-control" rows="1" id="backup-target-rsync-host">
+    </div>
+  </div>
+  <div class="form-group backup-target-rsync">
+    <label for="backup-target-rsync-path" class="col-sm-2 control-label">Path</label>
+    <div class="col-sm-8">
+        <input type="text" placeholder="/backups/{{hostname}}" class="form-control" rows="1" id="backup-target-rsync-path">
+    </div>
+  </div>
+  <div class="form-group backup-target-rsync">
+    <label for="backup-target-rsync-user" class="col-sm-2 control-label">Username</label>
+    <div class="col-sm-8">
+      <input type="text" class="form-control" rows="1" id="backup-target-rsync-user">
+    </div>
+  </div>
+  <div class="form-group backup-target-rsync">
+    <label for="ssh-pub-key" class="col-sm-2 control-label">Public SSH Key</label>
+    <div class="col-sm-8">
+      <input type="text" class="form-control" rows="1" id="ssh-pub-key" readonly>
+      <div class="small" style="margin-top: 2px">
+        Copy the Public SSH Key above, and paste it within the <tt>~/.ssh/authorized_keys</tt>
+        of target user on the backup server specified above. That way you'll enable secure and
+        passwordless authentication from your mail-in-a-box server and your backup server.
+      </div>
+    </div>
+  </div>
+  <!-- S3 BACKUP -->
   <div class="form-group backup-target-s3">
     <div class="col-sm-10 col-sm-offset-2">
       <p>Backups are stored in an Amazon Web Services S3 bucket. You must have an AWS account already.</p>
@@ -60,7 +104,8 @@
       <input type="text" class="form-control" rows="1" id="backup-target-pass">
     </div>
   </div>
-  <div class="form-group backup-target-local backup-target-s3">
+  <!-- Common -->
+  <div class="form-group backup-target-local backup-target-rsync backup-target-s3">
     <label for="min-age" class="col-sm-2 control-label">Days:</label>
     <div class="col-sm-8">
       <input type="number" class="form-control" rows="1" id="min-age">
@@ -92,7 +137,7 @@
 
 function toggle_form() {
   var target_type = $("#backup-target-type").val();
-  $(".backup-target-local, .backup-target-s3").hide();
+  $(".backup-target-local, .backup-target-rsync, .backup-target-s3").hide();
   $(".backup-target-" + target_type).show();
 }
 
@@ -160,28 +205,37 @@ function show_system_backup() {
 }
 
 function show_custom_backup() {
-    $(".backup-target-local, .backup-target-s3").hide();
+    $(".backup-target-local, .backup-target-rsync, .backup-target-s3").hide();
     api(
       "/system/backup/config",
       "GET",
       { },
       function(r) {
+        $("#backup-target-user").val(r.target_user);
+        $("#backup-target-pass").val(r.target_pass);
+        $("#min-age").val(r.min_age_in_days);
+        $(".backup-location").text(r.file_target_directory);
+        $(".backup-encpassword-file").text(r.enc_pw_file);
+        $("#ssh-pub-key").val(r.ssh_pub_key);
+
         if (r.target == "file://" + r.file_target_directory) {
           $("#backup-target-type").val("local");
         } else if (r.target == "off") {
           $("#backup-target-type").val("off");
+        } else if (r.target.substring(0, 8) == "rsync://") {
+          $("#backup-target-type").val("rsync");
+          var path = r.target.substring(8).split('//');
+          var [ user, host ] = path.shift().split('@');
+          $("#backup-target-rsync-user").val(user);
+          $("#backup-target-rsync-host").val(host);
+          $("#backup-target-rsync-path").val('/'+path);
         } else if (r.target.substring(0, 5) == "s3://") {
           $("#backup-target-type").val("s3");
-	  var hostpath = r.target.substring(5).split('/');
+          var hostpath = r.target.substring(5).split('/');
           var host = hostpath.shift();
           $("#backup-target-s3-host").val(host);
           $("#backup-target-s3-path").val(hostpath.join('/'));
         }
-        $("#backup-target-user").val(r.target_user);
-        $("#backup-target-pass").val(r.target_pass);
-        $("#min-age").val(r.min_age_in_days);
-        $('#backup-location').text(r.file_target_directory);
-        $('.backup-encpassword-file').text(r.enc_pw_file);
         toggle_form()
       })
 }
@@ -196,6 +250,12 @@ function set_custom_backup() {
     target = target_type;
   else if (target_type == "s3")
     target = "s3://" + $("#backup-target-s3-host").val() + "/" + $("#backup-target-s3-path").val();
+  else if (target_type == "rsync") {
+    target = "rsync://" + $("#backup-target-rsync-user").val() + "@" + $("#backup-target-rsync-host").val()
+                                                                + "/" + $("#backup-target-rsync-path").val();
+    target_user = '';
+  }
+
 
   var min_age = $("#min-age").val();
   api(

setup/system.sh

@@ -214,6 +214,12 @@ pollinate  -q -r
 
 # Between these two, we really ought to be all set.
 
+# We need an ssh key to store backups via rsync, if it doesn't exist create one
+if [ ! -f /root/.ssh/id_rsa_miab ]; then
+	echo 'Creating SSH key for backup…'
+	ssh-keygen -t rsa -b 2048 -a 100 -f /root/.ssh/id_rsa_miab -N '' -q
+fi
+
 # ### Package maintenance
 #
 # Allow apt to install system updates automatically every day.