On Mon, 10 Mar 2003, Tom Lane wrote:
> Lonni J Friedman <lfriedman@vasoftware.com> writes:
> > Well, it was over a year ago. basically what happened was that whenever i
> > set the value to be anything greater than 32, postmaster failed to start.
>
> Failed to start with what error message, exactly? It sounds to me like
i wish i remembered, but like i said, it was some time ago, so all i
recollect is the problem, not the details. sorry.
> you are describing a problem with unreasonably low kernel parameters,
> not with Postgres. (In particular I'd wonder what SEMMAX is on your
> system, though I suppose SHMMAX might be the issue too. See the Admin
> Guide concerning care and feeding of kernel parameters.)
i considered this, but the weird thing is that i know i didn't touch any
of the /proc bound kernel parms yet a recompile of postgres miraculously
resolved the problem.
>
> > bring it back down to a value less than or equal to 32, and it started up
> > fine. after recompiling postgresql with a hardcoded value of 64, then i
> > was able to increase it in postgresql.conf.
>
> Hardcoded value of what?
max_connections. once again, i dont' clearly remember the details. yea,i
know this doesn't shed any light on the problem. i just thought maybe
this was a known restriction. if you say its not any longer, i'll gladly
take your word on it. i'm no longer the admin of the box where this was
occuring last year.
thanks for your reply.
