> -----Original Message-----
> From: Lamar Owen [mailto:[EMAIL PROTECTED]]
> Sent: 21 September 1999 21:13
> To: Colin Price (EML)
> Cc: '[EMAIL PROTECTED]'
> Subject: Re: [ADMIN] [PostgreSQL] - Memory leek & duplicate postmaster
> processes
>
>
> "Colin Price (EML)" wrote:
> > $ more /etc/issue
> > Red Hat Linux release 5.2 (Apollo)
> > Kernel 2.2.1 on an i686
> > $ more /var/lib/pgsql/PG_VERSION
> > 6.4
>
> Aha! There are known issues with kernel 2.2.1. Either downgrade to
> kernel 2.0.3[68] or upgrade to 2.2.5+. Or, upgrade to RedHat 6.0.
>
> You can also download my 6.5.[12] RPMs, which will also help
> you upgrade
> your PostgreSQL databases. You are running the rpm version
> (/var/lib/pgsql is a dead giveaway).
>
Yeah - I've updated to
- Red Hat Linux release 6.0 (Hedwig),
- Kernel 2.2.5-22 on an i686
- PostgreSQL 6.5.
It seems fine .. except that the previous version
- Red Hat Linux release 5.2 (Apollo)
- Kernel 2.2.1 on an i686
- PosgreSQL 6.4
was fine for ~9 months before it started to leak. I hadn't changed any of config for
around
5 months. .... does this mean upgrading every ~9 months? Doesn't sound that bad I
suppose but
I really would like to get to the problem, not go round it.
> You can use netstat -a to find out what's connecting to any give
> unix-domain socket or internet socket.
>
> Unless clients are attempting to connect immediately on bootup,
> postmaster shouldn't fork a backend at all, until a client
> connection is established.
> What kind of clients do you have?
>
I have no established connections (ie, no one log's in). They only log in through a
Web interface (via TCP/IP)
Actually, how did you configure tcp/ip router/gateway table's. I have been 'playing'
but there is so much doc.
Again, any further advice to the above would be appreciated.
- Colin
************
