Re: TRUNCATE - timing of the return of disk space - caused by long-lived client?

[Alvaro Herrera <alvherre(at)commandprompt(dot)com>]

Vince Negri wrote:

> So I tried TRUNCATE on the table. It appeared to work  - in that the
> row count dropped to zero and the connected client was not disrupted,
> and "du" on the postgres data directory showed a fall. 
> 
> But the available disk space (reported by "df") did not fall. 

I think you shouldn't worry about it.  The file will eventually be
closed (maybe after a couple of checkpoints) and the space returned to
the filesystem.

FYI what TRUNCATE does is create a new, separate file for the table and
index storages.  The old one can still be open for a while, but it
should get detached not long after the transaction commits.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support