Thread: Configing 8 gig box.
Seems to be only using like 360 meg out of 7 gig free (odd thing is I did see some used swap 4k out of 1.9) with a bunch of users (this may be normal, but it is not going overly fast so thought I would ask).
Items I modified per commandprompt.coma nd watching this list etc.
shared_buffers = 24576
work_mem = 32768
max_fsm_pages = 100000
max_fsm_relations = 1500
fsync = true
wal_sync_method = open_sync
wal_buffers = 2048
checkpoint_segments = 100
effective_cache_size = 524288
default_statistics_target = 250
Any help is appreciated.
Joel Fradkin
Wazagua, Inc.
2520 Trailmate Dr
Sarasota, Florida 34243
Tel. 941-753-7111 ext 305
jfradkin@wazagua.com
www.wazagua.com
Powered by Wazagua
Providing you with the latest Web-based technology & advanced tools.
© 2004. WAZAGUA, Inc. All rights reserved. WAZAGUA, Inc
This email message is for the use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and delete and destroy all copies of the original message, including attachments.
On Mon, May 09, 2005 at 04:55:53PM -0400, Joel Fradkin wrote: > Seems to be only using like 360 meg out of 7 gig free (odd thing is I did > see some used swap 4k out of 1.9) with a bunch of users (this may be normal, > but it is not going overly fast so thought I would ask). This is perfectly normal. Each postgresql backend will only report memory usage roughly equal to shared_buffers plus the size of the code (16M or so?). If it's in the middle of a sort or vacuum, it will use more memory. It's not uncommon for modern OS's to swap out stuff that's not being used. They would rather have the memory available for disk caching, which is normally a good trade-off. For reference, on a 4G box running FreeBSD, there's currently 18M of swap used. Postgresql processes typically show 53M of total VM, with ~22M resident. This is with shared buffers of 2000. > Items I modified per commandprompt.coma nd watching this list etc. > > > > shared_buffers = 24576 > > work_mem = 32768 > > max_fsm_pages = 100000 > > max_fsm_relations = 1500 > > fsync = true > > wal_sync_method = open_sync > > wal_buffers = 2048 > > checkpoint_segments = 100 > > effective_cache_size = 524288 > > default_statistics_target = 250 > > > > Any help is appreciated. > > > > > > > > Joel Fradkin > > > > Wazagua, Inc. > 2520 Trailmate Dr > Sarasota, Florida 34243 > Tel. 941-753-7111 ext 305 > > > > jfradkin@wazagua.com > www.wazagua.com > Powered by Wazagua > Providing you with the latest Web-based technology & advanced tools. > C 2004. WAZAGUA, Inc. All rights reserved. WAZAGUA, Inc > This email message is for the use of the intended recipient(s) and may > contain confidential and privileged information. Any unauthorized review, > use, disclosure or distribution is prohibited. If you are not the intended > recipient, please contact the sender by reply email and delete and destroy > all copies of the original message, including attachments. > > > > > > > > -- Jim C. Nasby, Database Consultant decibel@decibel.org Give your computer some brain candy! www.distributed.net Team #1828 Windows: "Where do you want to go today?" Linux: "Where do you want to go tomorrow?" FreeBSD: "Are you guys coming, or what?"
"Jim C. Nasby" <decibel@decibel.org> writes: > On Mon, May 09, 2005 at 04:55:53PM -0400, Joel Fradkin wrote: >> Seems to be only using like 360 meg out of 7 gig free (odd thing is I did >> see some used swap 4k out of 1.9) with a bunch of users (this may be normal, >> but it is not going overly fast so thought I would ask). > This is perfectly normal. Each postgresql backend will only report > memory usage roughly equal to shared_buffers plus the size of the code > (16M or so?). If it's in the middle of a sort or vacuum, it will use > more memory. One thing to note is that depending on which Unix variant you are using, top may claim that any particular backend process is using the portion of shared memory that it's actually physically touched. This means that the claimed size of a backend process will grow as it runs (and randomly needs to touch pages that are in different slots of the shared-memory buffers) regardless of any actual objective growth in memory needs. regards, tom lane