Re: [HACKERS] Bug? relpages, reltuples resets to zero

> Hi,
>
>     there  seems  to be a problem with the relation statistics in
>     pg_class. Could someone explain why this happens?
>
>         doc=> vacuum;
>         VACUUM
>         doc=> select relname, relpages, reltuples from pg_class
>         doc-> where relname = 'doc_wordref';
>         relname    |relpages|reltuples
>         -----------+--------+---------
>         doc_wordref|    1099|   136027
>         (1 row)
>
>         -- ******** That's right
>
>         doc=> explain select distinct refpage from doc_wordref
>         doc-> where refword ~ '^a';
>         NOTICE:  QUERY PLAN:
>

Attached is a patch to vacuum.c I would like you to try.  I wonder if
dirty pages are somehow never getting written out because of vacuum's
use of transactions.


--
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026
*** ./backend/commands/vacuum.c.orig    Thu Oct 22 17:14:37 1998
--- ./backend/commands/vacuum.c    Thu Oct 22 17:15:35 1998
***************
*** 1875,1886 ****
          heap_close(sd);
      }

-     /* XXX -- after write, should invalidate relcache in other backends */
-     WriteNoReleaseBuffer(ItemPointerGetBlockNumber(&rtup->t_ctid));
-
      RelationInvalidateHeapTuple(rd, rtup);

!     ReleaseBuffer(buffer);
      heap_close(rd);
  }

--- 1875,1885 ----
          heap_close(sd);
      }

      RelationInvalidateHeapTuple(rd, rtup);

!     /* XXX -- after write, should invalidate relcache in other backends */
!     WriteBuffer(ItemPointerGetBlockNumber(&rtup->t_ctid));
!
      heap_close(rd);
  }