Обсуждение: bits & the end of the world

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I just got everyone's favorite error message:
FATAL 1:  my bits moved right off the end of the world!
        Recreate index pg_attribute_relid_attnum_index.

Except, of course, this is a system index.

The query was:

CREATE TEMP TABLE NextHopTemp (num     integer,
                               prev    integer,
                               settled boolean,
                               cost    integer);

Note also that this was inside a transaction.

I am running postgresql 7.0.2 from the RPM. Any ideas? :o

Ian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.1 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE5qb4wfn9ub9ZE1xoRAoYNAKDg4Vl3sYUfG87AZx4AnfAsgLWymQCfdLJK
wyEeK6dcSw1VFZrYOM/taZU=
=gL2K
-----END PGP SIGNATURE-----

-- Start of PGP signed section.
> I just got everyone's favorite error message:
> FATAL 1:  my bits moved right off the end of the world!
>         Recreate index pg_attribute_relid_attnum_index.

Yes, this has always been my favorite.

> Except, of course, this is a system index.
>
> The query was:
>
> CREATE TEMP TABLE NextHopTemp (num     integer,
>                                prev    integer,
>                                settled boolean,
>                                cost    integer);
>
> Note also that this was inside a transaction.
>
> I am running postgresql 7.0.2 from the RPM. Any ideas? :o

Sorry, no ideas.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@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

-- Start of PGP signed section.
> I just got everyone's favorite error message:
> FATAL 1:  my bits moved right off the end of the world!
>         Recreate index pg_attribute_relid_attnum_index.
>
> Except, of course, this is a system index.
>
> The query was:
>
> CREATE TEMP TABLE NextHopTemp (num     integer,
>                                prev    integer,
>                                settled boolean,
>                                cost    integer);
>
> Note also that this was inside a transaction.
>
> I am running postgresql 7.0.2 from the RPM. Any ideas? :o

I had this in my mailbox.  Because it is a system index, you may be
stuck.  You may need to use pg_upgrade.  This will allow you to
re-create the tables, and move the data files back into place, fixing
the problem.  If I remember correctly, you can try REINDEX, which is a
new command in 7.0.X for rebuilding system indexes in place.


> It is from the btree index code, nbtpage.c
>
>         if (P_RIGHTMOST(opaque))
>             elog(FATAL, "my bits moved right off the end of the
> world!");
>
> I would drop the index and recreate.  The wording of this error message
> is one of my personal favorites.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@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

>> I just got everyone's favorite error message:
>> FATAL 1:  my bits moved right off the end of the world!
>> Recreate index pg_attribute_relid_attnum_index.

>> Except, of course, this is a system index.

There's a REINDEX command now that you can use in standalone mode
to recover from something like this.

The procedure is arcane (translation: I don't know it) so check
the documentation and follow it carefully.

I've fixed a few bugs in btree for 7.1 and am optimistically hoping
that this message will not be heard from again after 7.1 is out...

            regards, tom lane