Hi all,
I'm having a problem with postgresql 7.1.2
running under solaris. I have the server
on one maching running. Occassionally,
all at once, I'll have about 200 clients
try to connect at once for a short time using
JDBC. Postgress will only accept 32 users
at a time (Solaris does not seem to like it
when I try to increase this number). If a
client can't get a connection, it waits for
a random number of seconds (between 1 and 10)
and then tries again until it gets a connection.
The problem: What seems to happen is that
one backend does all the work and the others
just sit and do nothing. When I run top,
I see one postgres process using about 1%
of the processor time and a large number
down the list doing nothing. Why won't
postgress kill the other backends and reuse
the connections? They are being closed
by the JDBC client.
Thanks,
Travis
--
---------------------------------------
| Travis Bauer | CS Grad Student | IU |
---------------------------------------
| http://www.cs.indiana.edu/~trbauer |
---------------------------------------
GPG Public Key:
http://www.cs.indiana.edu/~trbauer/publickey.html