On Mon, Jul 31, 2006 at 06:09:33PM -0400, Francisco Reyes wrote:
> Martijn van Oosterhout writes:
>
> >That's when you've reached the end of the table. The point is that
> >before then you'll have found the value of N that produces the error.
>
> Will be a while.. my little python script is doing under 10 selects/sec...
> and there are nearly 67 million records. :-(
Naturally you'd do a binary search. That'd take a maximum of about
log2(67 million) = about 26 scans.
Once you find it you should be able to identify the ctid. You may be
able to delete it.
> >It will rollback all pending transactions. The point is that it's
> >looking for information about transactions that were committed. This is
> >usually a memory or disk error.
>
> So, should it be safe to create the file and fill it up with 256K zeros?
For a certain value of "safe". It get the system running, but there's
no guarentees about the data in it...
> At the rate my script is going.. it's going to take a very long time to
> find out where the problem is. If I have a dump.. any usefull info I can
> take from the point the dump stopped?
That gives you the place in the file where it broke...
Have a nice day,
--
Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/
> From each according to his ability. To each according to his ability to litigate.
