Merge pull request #752 from enkore/feature-ignore-inode

Feature --ignore--inode, fixes #628

borg/archive.py

@@ -517,7 +517,7 @@ Number of files: {0.stats.nfiles}'''.format(
         self.add_item(item)
         return 'i'  # stdin
 
-    def process_file(self, path, st, cache):
+    def process_file(self, path, st, cache, ignore_inode=False):
         status = None
         safe_path = make_path_safe(path)
         # Is it a hard link?
@@ -533,7 +533,7 @@ Number of files: {0.stats.nfiles}'''.format(
                 self.hard_links[st.st_ino, st.st_dev] = safe_path
         path_hash = self.key.id_hash(os.path.join(self.cwd, path).encode('utf-8', 'surrogateescape'))
         first_run = not cache.files
-        ids = cache.file_known_and_unchanged(path_hash, st)
+        ids = cache.file_known_and_unchanged(path_hash, st, ignore_inode)
         if first_run:
             logger.info('processing files')
         chunks = None

borg/archiver.py

@@ -201,6 +201,7 @@ class Archiver:
 
         self.output_filter = args.output_filter
         self.output_list = args.output_list
+        self.ignore_inode = args.ignore_inode
         dry_run = args.dry_run
         t0 = datetime.utcnow()
         if not dry_run:
@@ -242,7 +243,7 @@ class Archiver:
         if stat.S_ISREG(st.st_mode) or read_special and not stat.S_ISDIR(st.st_mode):
             if not dry_run:
                 try:
-                    status = archive.process_file(path, st, cache)
+                    status = archive.process_file(path, st, cache, self.ignore_inode)
                 except OSError as e:
                     status = 'E'
                     self.print_warning('%s: %s', path, e)
@@ -965,6 +966,12 @@ class Archiver:
         traversing all paths specified. The archive will consume almost no disk space for
         files or parts of files that have already been stored in other archives.
 
+
+        To speed up pulling backups over sshfs and similar network file systems which do
+        not provide correct inode information the --ignore-inode flag can be used. This
+        potentially decreases reliability of change detection, while avoiding always reading
+        all files on these file systems.
+
         See the output of the "borg help patterns" command for more help on exclude patterns.
         """)
 
@@ -1020,6 +1027,9 @@ class Archiver:
                                type=ChunkerParams, default=CHUNKER_PARAMS,
                                metavar='CHUNK_MIN_EXP,CHUNK_MAX_EXP,HASH_MASK_BITS,HASH_WINDOW_SIZE',
                                help='specify the chunker parameters. default: %d,%d,%d,%d' % CHUNKER_PARAMS)
+        subparser.add_argument('--ignore-inode', dest='ignore_inode',
+                               action='store_true', default=False,
+                               help='ignore inode data in the file metadata cache used to detect unchanged files.')
         subparser.add_argument('-C', '--compression', dest='compression',
                                type=CompressionSpec, default=dict(name='none'), metavar='COMPRESSION',
                                help='select compression algorithm (and level): '

borg/cache.py

@@ -39,6 +39,12 @@ class Cache:
 
     def __init__(self, repository, key, manifest, path=None, sync=True, do_files=False, warn_if_unencrypted=True,
                  lock_wait=None):
+        """
+        :param do_files: use file metadata cache
+        :param warn_if_unencrypted: print warning if accessing unknown unencrypted repository
+        :param lock_wait: timeout for lock acquisition (None: return immediately if lock unavailable)
+        :param sync: do :meth:`.sync`
+        """
         self.lock = None
         self.timestamp = None
         self.lock = None
@@ -394,7 +400,7 @@ Chunk index:    {0.total_unique_chunks:20d} {0.total_chunks:20d}"""
             self.chunks[id] = (count - 1, size, csize)
             stats.update(-size, -csize, False)
 
-    def file_known_and_unchanged(self, path_hash, st):
+    def file_known_and_unchanged(self, path_hash, st, ignore_inode=False):
         if not (self.do_files and stat.S_ISREG(st.st_mode)):
             return None
         if self.files is None:
@@ -403,7 +409,8 @@ Chunk index:    {0.total_unique_chunks:20d} {0.total_chunks:20d}"""
         if not entry:
             return None
         entry = msgpack.unpackb(entry)
-        if entry[2] == st.st_size and bigint_to_int(entry[3]) == st.st_mtime_ns and entry[1] == st.st_ino:
+        if (entry[2] == st.st_size and bigint_to_int(entry[3]) == st.st_mtime_ns and
+                (ignore_inode or entry[1] == st.st_ino)):
             # reset entry age
             entry[0] = 0
             self.files[path_hash] = msgpack.packb(entry)