I'm running Sambar 7.0, PHP 5.2 and PostgreSQL 8.2 with WinXP. I built a database that's fed with form data, and created summaries with views and PHP scripts. I cloned the database two different ways. One way used pg_dump out of the existing and pg_restore into the new, the other method was to maually execute the SQL to create all the objects. Using the same summary script edited to point to the other database no matter how it was created takes more than 20 minutes to display. Running the queries in pgAdmin gives results in ms. I've searched all the logs and can find no errors, warnings or notices. Postgres and PHP documentation indicate there shouldn't be a problem. I've figured out the second database isn't scanning the indexes that were set up. This would explain the fact that the PHP script works, but is extremely slow. Why would PostGres use the indexes in one database, but not the other? I noticed that someone in another forum created two identical databases in a cluster and can connect to one but not the other. This doesn't bode well with me. Using pgAdmin III to execute the views in both databases, they run comparably, in the ms. Pulling a web summary that has a years worth of data, ~1500 rows of summarized data, comes up in less than 20 seconds. Doing so with the second and third identical databases takes 20 - 30 minutes. The PHP code is identical with the exception of the database name. Configuration settings appear to be global, and so far small changes have yielded some overall improvement, but have not closed the gap in performance between the databases. There's not much documentation that addresses multiple databases in a cluster, and what there is simply indicates they should run comparably. I'm finding no documentation at all of how to configure an individual database in a cluster. Why be able to build multiple databases if only the first one works correctly? No errors, no log entries.... just horrible performance.
15
