Christopher Schultz wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
David,
David Smith wrote:
| I'll bet money the added, idle postgres connections are just sleeping
| while they wait for work. Given idle connections contribute virtually
| no additional load, don't see his argument that idle connections
| contribute to a database overload. If the minor increase in overhead
| due to sleeping threads actually overloads the database, you need to let
| your boss know the server hardware is way too frail for production use
| and needs to be upgraded.
It could be a memory issue, and perhaps with lots of connections from
different machines (maybe a clustered application and a single database
instance)... lots of connections == lots of memory allocated on the
server, even when the connections are idle.
- -chris
Chris,
but as in many resource allocation issues, its not the quiescent count
that's important. If the OP has a *maximum* requirement for N x 100
connections then its unlikely that the system could be so constrained as
to worry about the limited number of idle connections. In other words,
if you've allocated enough resource for the maximum expected load, then
any less load should be easily accomadated. If you haven't you've
probably got a problem...
Regards
Alan
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEARECAAYFAkgWGZcACgkQ9CaO5/Lv0PCVVgCgiQHs4I8xJfbupkGXUHEBLgci
qmQAn3yNwIsnO4jzF/GobTZyzXp8fe17
=ma/c
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To start a new topic, e-mail: [email protected]
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
!DSPAM:48161a6671411738514357!
---------------------------------------------------------------------
To start a new topic, e-mail: [email protected]
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
