From: Michael Vrable <mvrable@cs.ucsd.edu>
Date: Thu, 9 Sep 2010 06:44:49 +0000 (-0700)
Subject: Improve cleaner: make sure new logs are written after existing ones.
X-Git-Url: https://git.vrable.net/?a=commitdiff_plain;h=c665095a0fb48ac1304f906eaca563ba8d01e6e8;p=bluesky.git

Improve cleaner: make sure new logs are written after existing ones.
---

diff --git a/cleaner/cleaner b/cleaner/cleaner
index 42aef01..fb3ce05 100755
--- a/cleaner/cleaner
+++ b/cleaner/cleaner
@@ -49,6 +49,13 @@ class FileBackend:
         fp.write(data)
         fp.close()
 
+    def loc_to_name(self, location):
+        return "log-%08d-%08d" % (location)
+
+    def name_to_loc(self, name):
+        m = re.match(r"^log-(\d+)-(\d+)$", name)
+        if m: return (int(m.group(1)), int(m.group(2)))
+
 class LogItem:
     """In-memory representation of a single item stored in a log file."""
 
@@ -91,18 +98,23 @@ class LogSegment:
 
     def close(self):
         data = ''.join(self.data)
-        filename = "log-%08d-%08d" % (self.location)
+        filename = self.backend.loc_to_name(self.location)
         print "Would write %d bytes of data to %s" % (len(data), filename)
         self.backend.write(filename, data)
 
 class LogDirectory:
     TARGET_SIZE = 4 << 20
 
-    def __init__(self, backend, dir, seq):
+    def __init__(self, backend, dir):
         self.backend = backend
         self.dir_num = dir
-        self.seq_num = seq
+        self.seq_num = 0
+        for logname in backend.list():
+            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)
         self.groups = {}
+        print "Starting sequence number is", self.seq_num
 
     def open_segment(self):
         seg = LogSegment(self.backend, (self.dir_num, self.seq_num))
@@ -182,7 +194,7 @@ def load_item(backend, location):
 
     The elements of the tuple are (directory, sequence, offset, size)."""
 
-    filename = "log-%08d-%08d" % (location[0], location[1])
+    filename = backend.loc_to_name((location[0], location[1]))
     data = backend.read(filename)[location[2] : location[2] + location[3]]
     item = parse_item(data)
     item.location = location
@@ -330,15 +342,15 @@ def run_cleaner(backend, inode_map, log):
     for (s, u) in sorted(inode_map.util.segments.items()):
         if float(u[1]) / u[0] < 0.99 and u[1] > 0:
             print "Should clean segment", s
-            m = re.match(r"^log-(\d+)-(\d+)$", s)
-            if m: inode_map.obsolete_segments.add((int(m.group(1)), int(m.group(2))))
+            loc = backend.name_to_loc(s)
+            if s: inode_map.obsolete_segments.add(loc)
 
     # Given that list of segments to clean, scan through those segments to find
     # data which is still live and mark relevant inodes as needing to be
     # rewritten.
     dirty_inodes = set()
     for s in inode_map.obsolete_segments:
-        filename = "log-%08d-%08d" % s
+        filename = backend.loc_to_name(s)
         print "Scanning", filename, "for live data"
         for item in parse_log(backend.read(filename), filename):
             if item.type in (ITEM_TYPE.DATA, ITEM_TYPE.INODE):
@@ -356,7 +368,7 @@ if __name__ == '__main__':
     imap.build(backend, chkpt)
     print chkpt
 
-    log_dir = LogDirectory(backend, 1, 0)
+    log_dir = LogDirectory(backend, 0)
     run_cleaner(backend, imap, log_dir)
     imap.write(backend, log_dir)
     log_dir.close_all()