Joe Lester <joe_lester(at)sweetwater(dot)com> writes:
> And it's common for the clients to stay running for days at a time. I'd
> say the average length of a connection is 3 days.
Uh-huh. Can you do anything to increase that?
Another possibility for slowing the leakage rate (pending a real fix)
is to decrease the size of your shared memory segment, ie, reduce
shared_buffers and the other shared-memory sizing parameters as much
as you can without killing performance.
regards, tom lane
In response to
pgsql-general by date
|Next:||From: Oliver Elphick||Date: 2004-08-20 20:10:15|
|Subject: Re: problem with postgresql-dump while upgrading to 7.4|
|Previous:||From: Bill Moran||Date: 2004-08-20 20:01:11|
|Subject: Re: Postgres filling up hard drive with swap files|