Hello,
Ich have a big performance problem. I'm running postgres 7.4.2 on Suse Linux 9.0 on a dual Xeon 3.0 GHz with 3 Gbyte Ram. In postgres.conf I'm using the defaults.
That's the place to start. See the guide at:
http://www.varlena.com/varlena/GeneralBits/Tidbits/perf.html
For your hardware, the default configuration settings are far too small. Oh, and you should upgrade to the latest 7.4 as soon as convenient.
> Filesystem is ext3 with writeback
journaling
I have 3 tables with ca 10 million entries with a gist index on GIS data and 5 tables with 10 million entries with an index on (timestamp,double,double). There are 10 tables with 1 million entries and index on int. and some smaller tables.
With 1 Gbyte Ram all went fine. Than I added a new table and it startet to swap. I added 2 Gbyte but the Problem is still there. The kswapd and kjournald are running nearly permanently.
If the system is swapping that's not likely to be due to PostgreSQL, especially on the default configuration settings.
The first time I do a query it takes very long. But the second time it goes a lot faster.
That's because the data is cached in RAM the second time.
Is postgres only using a certain amount of Ram for the indexes? But why my Ram is full then? Am I too short of Ram? Is the filesystem too slow?
What is "top" showing for memory usage? What does vmstat show for activity when you are having problems?
-- Richard Huxton Archonet Ltd
---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faqs/FAQ.html
