Tom Lane 写道:
>Neil Conway <neilc@xxxxxxxxxxx> writes:
>
>
>>Have you run ANALYZE recently? You might be running into the well-known
>>problem that hashed aggregation can consume an arbitrary amount of
>>memory -- posting the EXPLAIN for the query would confirm that.
>>
>>
>
>It would be useful to confirm whether this behavior is any different
>from 8.0 ...
>
>
>
Ok, rerun the query in another 8.0.3 box, it comes with 1G RAM and it's
linux,
without the 512M data segment limit (ulimit -d unlimited)。Load the data
and run vacuum analyze; then run the query, now the query return the result,
through top, we found that the postmaster process used 600+M RAM, the
enable_hashagg is ON.
wish that helps.
regards laser
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
