Re: clean up the temp space

I think there is a bug, though, in the 're-use' algorithm. It's hard to prove, but I reckon I have frequently seen situations where available extents (i.e. as previously allocated according to V$Sort_segment, not simply unallocated space in the tablespace) are NOT handed out to processes needing space. This was definitely the case with 7.3.3.5 and may still persist.

If people use the AUTOEXTEND feature on temporary tablespaces, this may allow the problem to escalate to ridiculous levels; so I would advise anyone with 'rapidly growing' temp tablespaces to make sure that they are NOT allowed to autoextend; this may solve the problem.

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

Howard J. Rogers wrote in message <39b0ae91_at_news.iprimus.com.au>...


>

>

>Can't comment on whether it's a bug or not, but proper temporary tablespace

>is supposed to sit there fully used.  Extents are never deallocated -not

>until you do a shutdown anyway.  They are simply re-used by other sorts.

 In


>permanent tablespace, extents are dropped at the end of the initiating sort

>(and hence have to be re-created by any subsequent sorts -which is

>comparitively slow).

>