Thread: Documentation bug at: current/sql-vacuum.html, Notes

Documentation bug at: current/sql-vacuum.html, Notes

From
Christophe Pettus
Date:
Isn't this backwards?

> If index cleanup is not performed regularly, performance may suffer, because as the table is modified indexes will
accumulatedead tuples and the table itself will accumulate dead line pointers that cannot be removed until index
cleanupis completed. 

Indexes have line pointers, tables have tuples, so it should read:

> If index cleanup is not performed regularly, performance may suffer, because as the table is modified indexes will
accumulatedead line pointers and the table itself will accumulate dead tuples that cannot be removed until index
cleanupis completed. 


Re: Documentation bug at: current/sql-vacuum.html, Notes

From
Nathan Bossart
Date:
On Wed, Jul 26, 2023 at 10:39:41AM -0700, Christophe Pettus wrote:
> Isn't this backwards?
> 
>> If index cleanup is not performed regularly, performance may suffer, because as the table is modified indexes will
accumulatedead tuples and the table itself will accumulate dead line pointers that cannot be removed until index
cleanupis completed.
 
> 
> Indexes have line pointers, tables have tuples, so it should read:
> 
>> If index cleanup is not performed regularly, performance may suffer, because as the table is modified indexes will
accumulatedead line pointers and the table itself will accumulate dead tuples that cannot be removed until index
cleanupis completed.
 

I believe the current phrasing is correct.  VACUUM will be able to reclaim
the tuple space in the heap, but the line pointers must remain until their
corresponding index tuples are removed.

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com



Re: Documentation bug at: current/sql-vacuum.html, Notes

From
Christophe Pettus
Date:

> On Jul 26, 2023, at 10:55, Nathan Bossart <nathandbossart@gmail.com> wrote:
> 
> 
> I believe the current phrasing is correct.  VACUUM will be able to reclaim
> the tuple space in the heap, but the line pointers must remain until their
> corresponding index tuples are removed.

Ah, right, thanks.