X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=cleaner%2Fcleaner;h=675819bad7845f016454f9a6c70a98d40c9258dc;hb=4a22b094a5dbd2aabc101ee5a6579b577b6e614c;hp=c4b1222cbea619460f733d2dc78642345bb63237;hpb=2223a4a4ec1ea91f30e345179cea751c022e5b99;p=bluesky.git diff --git a/cleaner/cleaner b/cleaner/cleaner index c4b1222..675819b 100755 --- a/cleaner/cleaner +++ b/cleaner/cleaner @@ -19,6 +19,8 @@ HEADER_MAGIC1 = 'AgI-' # Unencrypted data HEADER_MAGIC2 = 'AgI=' # Encrypted data HEADER_SIZE = struct.calcsize(HEADER_FORMAT) +CHECKPOINT_MAGIC = struct.pack(' 0: fp.seek(offset) - if legnth is None: + if length is None: return fp.read() else: return fp.read(length) @@ -93,9 +96,10 @@ class S3Backend: self.conn = boto.connect_s3(is_secure=False) self.bucket = self.conn.get_bucket(self.bucket_name) - def list(self): + def list(self, directory=0): files = [] - for k in self.bucket.list(self.path + 'log-'): + prefix = "log-%08d-" % (directory,) + for k in self.bucket.list(self.path + prefix): files.append((k.key, k.size)) return files @@ -208,7 +212,8 @@ class LogDirectory: self.backend = backend self.dir_num = dir self.seq_num = 0 - for logname in backend.list(): + for logname in backend.list(dir): + print "Old log file:", logname loc = backend.name_to_loc(logname[0]) if loc is not None and loc[0] == dir: self.seq_num = max(self.seq_num, loc[1] + 1) @@ -241,7 +246,7 @@ class UtilizationTracker: def __init__(self, backend): self.segments = {} - for (segment, size) in backend.list(): + for (segment, size) in backend.list(0) + backend.list(1): self.segments[segment] = [size, 0] def add_item(self, item): @@ -327,8 +332,8 @@ def parse_log(data, location=None): if item is not None: yield item offset += size -def load_checkpoint_record(backend): - for (log, size) in reversed(backend.list()): +def load_checkpoint_record(backend, directory=0): + for (log, size) in reversed(backend.list(directory)): for item in reversed(list(parse_log(backend.read(log), log))): print item if item.type == ITEM_TYPE.CHECKPOINT: @@ -343,6 +348,7 @@ class InodeMap: This will also build up information about segment utilization.""" + self.version_vector = {} self.checkpoint_record = checkpoint_record util = UtilizationTracker(backend) @@ -350,9 +356,25 @@ class InodeMap: inodes = {} self.obsolete_segments = set() + data = checkpoint_record.data + if not data.startswith(CHECKPOINT_MAGIC): + raise ValueError, "Invalid checkpoint record!" + data = data[len(CHECKPOINT_MAGIC):] + (vvlen,) = struct.unpack_from("