On Tue, Dec 07, 2004 at 12:02:13PM +1100, Neil Conway wrote: > On Mon, 2004-12-06 at 19:37 -0500, Paul Tillotson wrote: > > I seem to remember hearing that the memory limit on certain operations, > > such as sorts, is not "enforced" (may the hackers correct me if I am > > wrong); rather, the planner estimates how much a sort might take by > > looking at the statistics for a table.
> AFAIK this is not the case. AFAIK this is indeed the case with hashed aggregation, which uses the sort_mem (work_mem) parameter to control its operation, but for which it is not a hard limit. I concur however that multiple concurrent sorts may consume more memory than the limit specified for one sort. (Just last week I saw a server running with sort_mem set to 800 MB ... no wonder the server went belly up every day at 3.00am, exactly when a lot of reports were being generated) -- Alvaro Herrera (<[EMAIL PROTECTED]>) "Acepta los honores y aplausos y perderás tu libertad" ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match