lorry: backup .git in case of error

Backup the git repository before running mirror and if any part
of mirroring fails, restore the .git repository to its previous
state and keep the failed mirror for debugging.

1 files changed, 83 insertions, 7 deletions

diff --git a/lorry b/lorry
index 7135cea..b13709a 100755
--- a/lorry
+++ b/lorry
@@ -22,6 +22,8 @@ import os
 import urllib2
 import string
 import sys
+from datetime import datetime
+import shutil
 
 __version__ = '0.0'
 
@@ -148,18 +150,92 @@ class Lorry(cliapp.Application):
         if not os.path.exists(dirname):
             os.mkdir(dirname)
         gitdir = os.path.join(dirname, 'git')
-        table[vcstype](name, dirname, gitdir, spec)
-        if self.settings['repack']:
-            self.progress('.. repacking %s git repository' % name)
-            self.run_program(['git', 'repack', '-a', '-d', '--depth=250',
-                              '--window=250'], cwd=gitdir)
-        self.bundle(name, gitdir)
+        backupdir = self.backup_gitdir(name, dirname, gitdir)
+        try:
+            table[vcstype](name, dirname, gitdir, spec)
+            if self.settings['repack']:
+                self.progress('.. repacking %s git repository' % name)
+                self.run_program(['git', 'repack', '-a', '-d', '--depth=250',
+                                  '--window=250'], cwd=gitdir)
+            self.bundle(name, gitdir)
+        except:
+            if backupdir is not None:
+                faildir = self.save_failgit(name, dirname, gitdir)
+                self.restore_backup(name, dirname, gitdir, backupdir)
+                self.output.write('Mirror of %s failed, state before mirror '
+                                  'is saved at %s and state after mirror is '
+                                  'saved at %s\n' % (name, backupdir, faildir))
+                logging.debug('Mirror of %s failed, state before mirror '
+                              'is saved at %s and state after mirror is '
+                              'saved at %s\n', name, backupdir, faildir)
+            raise
+
         if not self.settings['pull-only']:
             if 'refspecs' in spec:
                 self.push_to_mirror_server(gitdir, spec['refspecs'])
             else:
                 self.push_to_mirror_server(gitdir)
- 
+
+        if backupdir is not None:
+            self.progress('.. removing %s git repository backup' % name)
+            shutil.rmtree(backupdir)
+
+    def restore_backup(self, name, dirname, gitdir, backupdir):
+        self.progress('.. restoring %s good git repository' % name)
+        dotgit = os.path.join(gitdir, '.git')
+        if not os.path.exists(dotgit):
+            dotgit = gitdir
+        shutil.rmtree(dotgit)
+        self.copy_gitdir(backupdir, dotgit)
+
+    def save_failgit(self, name, dirname, gitdir):
+        self.progress('.. saving failed %s mirror git repository' % name)
+        dotgit = os.path.join(gitdir, '.git')
+        if not os.path.exists(dotgit):
+            dotgit = gitdir
+        time = datetime.now().strftime('%F-%T')
+        backupdir = os.path.join(dirname, "git-post-fail-%s" % time)
+        return self.copy_gitdir(dotgit, backupdir)
+
+    def backup_gitdir(self, name, dirname, gitdir):
+        dotgit = os.path.join(gitdir, '.git')
+        if not os.path.exists(dotgit):
+            dotgit = gitdir
+        time = datetime.now().strftime('%F-%T')
+        backupdir = os.path.join(dirname, "git-pre-update-%s" % time)
+        self.progress('.. backing up %s git repository to %s' %
+                      (name, backupdir))
+        return self.copy_gitdir(dotgit, backupdir)
+
+    def copy_gitdir(self, source, dest):
+        if not os.path.exists(source):
+            return None
+
+        # copy everything except the objects dir
+        def ignoreobjects(dirname, filenames):
+            if dirname.endswith(source):
+                return ['objects']
+            return []
+        shutil.copytree(source, dest, ignore=ignoreobjects)
+
+        # hardlink the objects
+        sourceobjects = os.path.join(source, 'objects')
+        assert os.path.exists(sourceobjects), "No source objects"
+        objectspath = os.path.join(dest, 'objects')
+        os.mkdir(objectspath)
+        for dirpath, dirnames, filenames in os.walk(objectspath):
+            assert dirpath.startswith(objectspath)
+            # strip objectspath and / from relpath
+            relpath = dirpath[len(objectspath)+1:]
+            for dirname in dirnames:
+                os.mkdir(os.path.join(dest, relpath, dirname))
+            for filename in filenames:
+                assert os.path.exists(os.path.join(dest, relpath))
+                os.link(os.path.join(dirpath, filename),
+                        os.path.join(dest, relpath, filename))
+        return dest
+
+
     def add_remote(self, name, gitdir):
 
         ''' Add mirror server as a remote to gitdir.