what's wrong with postgres 8

Поиск
Список
Период
Сортировка
От vinita bansal
Тема what's wrong with postgres 8
Дата
Msg-id BAY20-F1566FF6053817E1A0E58EECB460@phx.gbl
обсуждение исходный текст
Список pgsql-general
Hi,

I have a 64 bit Linux box with 32GB RAM and disk space approx 450 GB.
I am running a benchmark using filesystem type ext3 with data=writeback both
on Postgres 7 and 8. Time taken for benchmark run is very less compared to
the one on Postgres 8. I am esentially trying to get performance improvement
by using data=writeback. If I do not use data=writeback both for Postgres 7
and 8, I am getting almost the same numbers for benchmark run. Why is this
happening. why data=writeback is giving good results for Postgres7 and not
for Postgres 8??

I am using the following settings both for Postgres 7 and 8:

shared_buffers = 3000
work_mem = 100000
maintenance_work_mem = 100000
max_fsm_pages = 200000
bgwriter_percent = 0
wal_buffers = 64
checkpoint_segments = 256
checkpoint_timeout = 1800
effective_cache_size = 250000
random_page_cost = 2
stats_start_collector = true
stats_command_string = true
stats_row_level = true

Also, I would like to know
        - if using a Postgres 7 database dump to restore database on
Postgres 8 can cause this
        - using Postgres 7 jdbc driver for Postgres 8 can lead to a time
performance degradation (I am using postgres 7 driver now since using
Postgres 8 driver I am getting an error
          java.sql.SQLException: ERROR: operator does not exist: smallint =
boolean.

FYI: I have runstats on all tables both for Postgres 7 and 8 after doing a
restore.

Regards,
Vinita Bansal

_________________________________________________________________
News, views and gossip. http://www.msn.co.in/Cinema/ Get it all at MSN
Cinema!


В списке pgsql-general по дате отправления:

Предыдущее
От: Peter Eisentraut
Дата:
Сообщение: Re: limited disk space
Следующее
От: josue
Дата:
Сообщение: Re: Upgrade data