Re: index problem - Mailing list pgsql-general

From Lincoln Yeoh
Subject Re: index problem
Date
Msg-id 3.0.5.32.20000605151542.008c2880@pop.mecomb.po.my
Whole thread Raw
In response to Re: index problem  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: index problem  (Marcin Inkielman <marn@wsisiz.edu.pl>)
List pgsql-general
At 06:13 PM 01-06-2000 -0400, Tom Lane wrote:
>Marcin Inkielman <marn@wsisiz.edu.pl> writes:
>> I created an index using pgaccess rescently. the name of the index was
>> long:
>> "oceny_stud_numer_albumu_protokoloceny_stud"
>> now i am unable to vacuum my database.
>
>Oh dear :-( ... it seems that when you quote an identifier, the system
>forgets to make sure that it's truncated to no more than 31 characters.
>You've got a corrupted pg_class entry now for that index.
>
>> my question is:
>> ~~~~~~~~~~~~~~~
>> how may i delete this index in my original database???
>
>Dropping the table that the index is on should work.  Hopefully
>restoring just the one table is better than restoring your whole DB.

If dropping the index is not possible, how about the good ol filesystem rm?

I used to use that to deal with drop table; rollback;

The table was there but not quite there ;). So I had to rm it.

How about hexediting the relevant files. I did that once to fix an
application which stored its code in an Oracle DB. The database wouldn't
allow us to access the vendor's code even though we knew where the bug was,
so I just stopped the database engine, backed up the 200MB database file,
hexedited it and fixed it :).

That's the sort of thing you sometimes have to do with closed source apps...

Cheerio,
Link.


pgsql-general by date:

Previous
From: Haroldo Stenger
Date:
Subject: Re: Can I prevent a sequence from being updated when a transaction fail?
Next
From: Lincoln Yeoh
Date:
Subject: Re: query optimiser changes 6.5->7.0