Don't clear btpo_cycleid during _bt_vacuum_one_page.

tglsfdc · tglsfdc · commit 593a9631a794 · 2012-02-21T15:03:36.000-05:00
When "vacuuming" a single btree page by removing LP_DEAD tuples, we are not actually within a vacuum operation, but rather in an ordinary insertion process that could well be running concurrently with a vacuum. So clearing the cycleid is incorrect, and could cause the concurrent vacuum to miss removing tuples that it needs to remove. This is a longstanding bug introduced by commit e628464 of 2006-07-25. I believe it explains Maxim Boguk's recent report of index corruption, and probably some other previously unexplained reports. In 9.0 and up this is a one-line fix; before that we need to introduce a flag to tell _bt_delitems what to do.
diff --git a/src/backend/access/nbtree/nbtpage.c b/src/backend/access/nbtree/nbtpage.c
@@ -841,11 +841,9 @@ _bt_delitems_delete(Relation rel, Buffer buf,
 	PageIndexMultiDelete(page, itemnos, nitems);
 
 	/*
-	 * We can clear the vacuum cycle ID since this page has certainly been
-	 * processed by the current vacuum scan.
+	 * Unlike _bt_delitems_vacuum, we *must not* clear the vacuum cycle ID,
+	 * because this is not called by VACUUM.
 	 */
-	opaque = (BTPageOpaque) PageGetSpecialPointer(page);
-	opaque->btpo_cycleid = 0;
 
 	/*
 	 * Mark the page as not containing any LP_DEAD items.  This is not
@@ -854,6 +852,7 @@ _bt_delitems_delete(Relation rel, Buffer buf,
 	 * true and it doesn't seem worth an additional page scan to check it.
 	 * Remember that BTP_HAS_GARBAGE is only a hint anyway.
 	 */
+	opaque = (BTPageOpaque) PageGetSpecialPointer(page);
 	opaque->btpo_flags &= ~BTP_HAS_GARBAGE;
 
 	MarkBufferDirty(buf);
