From: Michael Vrable <mvrable@cs.ucsd.edu>
Date: Wed, 13 Feb 2008 01:05:33 +0000 (-0800)
Subject: Slight tweaks to the local database to improve cleaning procedures.
X-Git-Url: https://git.vrable.net/?a=commitdiff_plain;h=030ddd80e04aa65738bf21557dd541bec757de5b;p=cumulus.git

Slight tweaks to the local database to improve cleaning procedures.

In addition to marking objects in cleaned segments, mark the segment itself
as cleaned.
---

diff --git a/contrib/upgrade0.6-localdb.sql b/contrib/upgrade0.6-localdb.sql
index 7256f9f..ce8797c 100644
--- a/contrib/upgrade0.6-localdb.sql
+++ b/contrib/upgrade0.6-localdb.sql
@@ -15,7 +15,8 @@ create table segments_new (
     path text,
     checksum text,
     mtime real,
-    size integer
+    size integer,
+    expire_time integer
 );
 
 create table segments_used (
@@ -35,7 +36,7 @@ update snapshots set intent = 1;
 -- Compute the size of each of the segments, if possible, based on our
 -- knowledge of the objects stored in them.
 insert into segments_new
-select segmentid, segment, path, checksum, mtime, size
+select segmentid, segment, path, checksum, mtime, size, null as expire_time
 from
     (select segmentid, segment, path, checksum from segments)
 left join
diff --git a/lbs.py b/lbs.py
index 5213b89..4224cf9 100644
--- a/lbs.py
+++ b/lbs.py
@@ -483,6 +483,13 @@ class LocalDatabase:
 
         cur = self.cursor()
 
+        # Find the id of the last snapshot to be created.  This is used for
+        # measuring time in a way: we record this value in each segment we
+        # expire on this run, and then on a future run can tell if there have
+        # been intervening backups made.
+        cur.execute("select max(snapshotid) from snapshots")
+        last_snapshotid = cur.fetchone()[0]
+
         # Get the list of old snapshots for this scheme.  Delete all the old
         # ones.  Rules for what to keep:
         #   - Always keep the most recent snapshot.
@@ -523,12 +530,16 @@ class LocalDatabase:
         cur.execute("""delete from segments where segmentid not in
                            (select segmentid from segments_used)""")
 
-        # Finally, delete objects contained in non-existent segments.  We can't
-        # simply delete unused objects, since we use the set of unused objects
-        # to determine the used/free ratio of segments.
+        # Delete unused objects in the block_index table.  By "unused", we mean
+        # any object which was stored in a segment which has been deleted, and
+        # any object in a segment which was marked for cleaning and has had
+        # cleaning performed already (the expired time is less than the current
+        # largest snapshot id).
         cur.execute("""delete from block_index
-                       where segmentid not in
-                           (select segmentid from segments)""")
+                       where segmentid not in (select segmentid from segments)
+                          or segmentid in (select segmentid from segments
+                                           where expire_time < ?)""",
+                    (last_snapshotid,))
 
     # Segment cleaning.
     class SegmentInfo(Struct): pass
@@ -604,7 +615,11 @@ class LocalDatabase:
             raise TypeError("Invalid segment: %s, must be of type int or SegmentInfo, not %s" % (segment, type(segment)))
 
         cur = self.cursor()
-        cur.execute("update block_index set expired = 1 where segmentid = ?",
+        cur.execute("select max(snapshotid) from snapshots")
+        last_snapshotid = cur.fetchone()[0]
+        cur.execute("update segments set expire_time = ? where segmentid = ?",
+                    (last_snapshotid, id))
+        cur.execute("update block_index set expired = 0 where segmentid = ?",
                     (id,))
 
     def balance_expired_objects(self):
diff --git a/schema.sql b/schema.sql
index 8529e32..f406d91 100644
--- a/schema.sql
+++ b/schema.sql
@@ -19,7 +19,8 @@ create table segments (
     path text,
     checksum text,
     mtime real,
-    size integer
+    size integer,
+    expire_time integer         -- snapshotid of latest snapshot when expired
 );
 
 -- Index of all blocks which have been stored, by checksum.