Avoid changing an index's indcheckxmin horizon during REINDEX.

There can never be a need to push the indcheckxmin horizon forward, since
any HOT chains that are actually broken with respect to the index must
pre-date its original creation.  So we can just avoid changing pg_index
altogether during a REINDEX operation.

This offers a cleaner solution than my previous patch for the problem
found a few days ago that we mustn't try to update pg_index while we are
reindexing it.  System catalog indexes will always be created with
indcheckxmin = false during initdb, and with this modified code we should
never try to change their pg_index entries.  This avoids special-casing
system catalogs as the former patch did, and should provide a performance
benefit for many cases where REINDEX formerly caused an index to be
considered unusable for a short time.

Back-patch to 8.3 to cover all versions containing HOT.  Note that this
patch changes the API for index_build(), but I believe it is unlikely that
any add-on code is calling that directly.

1 files changed, 6 insertions, 0 deletions

diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c
index 10ec32c05de..ff228b7d531 100644
--- a/src/backend/commands/cluster.c
+++ b/src/backend/commands/cluster.c
@@ -1398,6 +1398,12 @@ finish_heap_swap(Oid OIDOldHeap, Oid OIDNewHeap,
 	 * advantage to the other order anyway because this is all transactional,
 	 * so no chance to reclaim disk space before commit.  We do not need a
 	 * final CommandCounterIncrement() because reindex_relation does it.
+	 *
+	 * Note: because index_build is called via reindex_relation, it will never
+	 * set indcheckxmin true for the indexes.  This is OK even though in some
+	 * sense we are building new indexes rather than rebuilding existing ones,
+	 * because the new heap won't contain any HOT chains at all, let alone
+	 * broken ones, so it can't be necessary to set indcheckxmin.
 	 */
 	reindex_flags = REINDEX_REL_SUPPRESS_INDEX_USE;
 	if (check_constraints)