Re: vacuum won't fix tx wraparound problem

Yeah I did wait long enough for the vacuum to finish. I did consider the prepared_transactions issue but I don't think we are using those. I'll look down that path though since I could be wrong about that.

On a related note I thought in 8.4 a successive vacuum would not take as long as the prior since it "knows where it left off". It doesn't seem to be working like that when running vacuum in a standalone instance; it takes just as long each time, 3-4 hours.

thanks for all your help,
Gene

On Jun 27, 2010, at 10:30 AM, Tom Lane wrote:

> Bill Moran <wmoran(at)potentialtech(dot)com> writes:
>> In response to Gene Hart <genekhart(at)gmail(dot)com>:
>>> PostgreSQL stand-alone backend 8.4.4
>>> backend> vacuum
>>> backend> ^D^D
>>> exit
>
>> Am I reading this wrong or did you not bother to allow the vacuum to finish?
>> Considering there's no command terminator (;) on the vacuum command, it's
>> unlikely that it ever actually started to do anything.
>
> No, Gene did it right --- standalone backends have a different command-line
> syntax. (I assume also that he observed a suitably long delay before
> the second backend> prompt came up...)
>
> I think Scott's idea of ancient prepared transactions is probably the
> most likely bet. Roll those back and then vacuum and you'll be OK.
>
> regards, tom lane