Thread: Vacuum performance question
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 My database is approximately 8gb in size. My application is 24/7. I'm constantly cycling through data. I analyze every 15 minutes but I'm noticing that during a vacuum the database becomes extremely sluggish. In some cases the operation of my software goes from 1-3 second runtime to 300+ seconds during a vacuum. Should I expect this with a vacuum? I've done reading online where people say they see only a 10% decrease in speed. Is this supposed to be the norm? I've allocated 196MB of RAM to vacuums. The system is a dual P4-2.4ghz w/ 1.5 gig of RAM w/ 36 gig of RAID mirrored disk. I'm runing 7.3.2 and I am upgrading to 7.3.3 next week. Please advise. - -- Jeremy M. Guthrie Systems Engineer Berbee 5520 Research Park Dr. Madison, WI 53711 Phone: 608-298-1061 Berbee...Decade 1. 1993-2003 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQE/O80UqtjaBHGZBeURAqmLAJ9kxh0DyaZm3pAP77XGlDTq5JcsowCfeqpC 36SjIo5XW44bEkmHnbwXXBQ= =oozQ -----END PGP SIGNATURE-----
> I've allocated 196MB of RAM to vacuums. I would be willing to bet that you have kicked the system into swap because of this. Hence the large decrease in speed. Try sliding back to 32MB for vacuum. A ton more ram doesn't really help it all that much.
Attachment
"Jeremy M. Guthrie" <jeremy.guthrie@berbee.com> writes:
> In some cases the operation of my software goes from 1-3 second runtime to
> 300+ seconds during a vacuum.
That seems like a lot. I concur with the nearby recommendation to
reduce vacuum_mem, but I think there may be another problem. You should
watch top, iostat, vmstat during a vacuum to try to see what resource is
getting saturated.
> I'm runing 7.3.2 and I am upgrading to 7.3.3 next week.
7.3.4, please. 7.3.3 has at least one nasty bug.
regards, tom lane