Re: please please please PLEASE help! - Mailing list pgsql-admin
From | Rob Bamber |
---|---|
Subject | Re: please please please PLEASE help! |
Date | |
Msg-id | NJEBKAGMPKNPDHGFFJJDKEJLCHAA.rbamber@openworld.org Whole thread Raw |
In response to | Re: please please please PLEASE help! (Kris Kiger <kris@musicrebellion.com>) |
Responses |
Re: please please please PLEASE help!
Re: please please please PLEASE help! |
List | pgsql-admin |
Another thought - We had a similar issue recently. Our support guys dropped the database and then rebuilt it from a dump file. The size of the data directory went down from 12GB to less than 2GB. According to the sys ad that did the work postgres is not very good a reclaiming disk space after large quantities of tuples are deleted over time. HTH -----Original Message----- From: pgsql-admin-owner@postgresql.org [mailto:pgsql-admin-owner@postgresql.org]On Behalf Of Kris Kiger Sent: 29 July 2004 16:17 To: pgsql-admin@postgresql.org Subject: Re: [ADMIN] please please please PLEASE help! Steve, Are the three SCSI drives raided? If so, I would move the operating database to that machine. From what I understand it can be a real hassle to setup/maintain sym-linked tables. If you don't have the option to move the database, I would explore one of these two routes: Route one is replication. Slony1 was just released to the open source community. Replicate the data to the new machine, then distribute the database reads between the master and the slave. Route two is symlink database logs/txlogs/etc to the new machine. It takes extra write needs off of your main machine. Not as effective as route 1, I would say, but it would definately lessen the load. Kris >Hi, > >I've asked this question a couple of times before on this forum but no >one seems to be nice enough to point me to the right direction or help >me out with any information, if possible. Please help me out with this >because this is a very serious issue for me and I need to learn more >about this. And here it is again: > >I've been running postgres on my server for over a year now and the >tables have become huge. I have 3 tables that have data over 10GB each >and these tables are read very very frequently. In fact, heavy searches >on these tables are expected every 2 to 3 minutes. This unfortunately >gives a very poor response time to the end user and so I'm looking at >other alternatives now. > >Currently, the postgresql installation is on a single disk and so all >the tables have their data read from a single disk. Searching on >different tables by multiple users at the same time results in very slow >searches, as it's mainly dependant on the spindle speed. I recently >gained access to another server which has 3 SCSI disks. I know there is >a way to mirror the tables across the three different disks but I'm not >sure if it's as easy as symlinking the files (WAL files only?) across. >Can anyone please tell me what to do here and how to harness the power >of the three SCSI drives that I have. Which files in the data directory >need to be moved? Is this safe? Can backups etc be easily done? Any >information will be greatly appreciated. Thank you, > > >Steve > > > ---------------------------(end of broadcast)--------------------------- TIP 8: explain analyze is your friend
pgsql-admin by date: