Re: Forcing connections closed

If the system in question is ...

1. a Linux or *BSD AND
2. has ipchains/iptables/whatever firewall software installed AND
3. the connections in question to be killed are coming in via
   port 5432/tcp

then I would think you could temporarily change the firewall configuration
on the system such that all traffic from the remote IP, or even localhost,
on 5432/tcp gets dropped and possibly the connection should drop as well.
If it doesn't then you may additionally tweak your tcp_keepalive_time to
0 before changing the firewall rule.  Just remember to drop that firewall
rule and set the tcp_keepalive_time back when you're done.

Just random thoughts.

Greg


-----Original Message-----
From: Juan Miguel
To: PGSQL Admin
Sent: 2/1/04 4:47 PM
Subject: Re: [ADMIN] Forcing connections closed


>'kill -9' PID will force all backends to flush cache for no good
reason.
>
>Note that on most unixes, a plain kill PID will send the term signal,
>which tells the process to poltely shutdown and release resources.
kill
>-9 is like using a sledge hammer to swat a fly for this.
>
>
>---------------------------(end of
broadcast)---------------------------
>TIP 4: Don't 'kill -9' the postmaster
>
>
>
;-) Yes, you are on the right. "kill -9", is the last try, but often you

have to do "kill -9" for killing a procces, even sometimes, 'kill -9'
does not work (often when the process is acceding to a hardware device -

a CD for example -).



---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faqs/FAQ.html