Re: Need help configuring Piranha. Have only a few days left to getthis done.

Hi Edward,

> [SMTP stuff]

You could have snipped it yourself as a courtesy to the list readers.

> The documentation states that you can use a script for the send,
> but it never reads it.

Did you have a look at the logs ? If there's nothing relevant, you might
restart pulse with -v and look again. IT works fine for me, so it should
also for you.

Here is what I have :
failover jonas {
        active = 1
        address = 172.22.40.103 eth0:3
        vip_nmask = 255.255.240.0
        port = 1099
        send_program = "/etc/sysconfig/ha/monitor.jonas.sh"
        expect = "OK"
        timeout = 45
        start_cmd = "/sbin/service jonas restart"
        stop_cmd = "/sbin/service jonas stop"
}

[sbt@ssca4010 sbt]$ cat /etc/sysconfig/ha/monitor.jonas.sh
#! /bin/sh
. /etc/profile.d/jonas.sh
$JONAS_ROOT/bin/unix/JonasAdmin -e 2>&1 | grep -q unreachable && echo
FAIL || echo OK

> I verified this by putting a mail command at the very beginning.
> It isn't a permissions issue as I su'ed into the postgres account
> and was able to run the scripts.

Watch out ! your script will be run as the same user pulse is, this is
generally root !

> Here is a copy of the lvs.cf from the database server.
>      vip_nmask = 255.255.255.0

Eventhough you could be right, 255.0.0.0 is the usual netmask for a
10.something network.

>      send_program = "/scripts/check_pgsql_alive"

Can you send it ?

> I had port 5432 in there, but took it out to test that also.

As you can see, I still have the port ... just in case :)

> [LVS config]
> network = nat

I'm assuming a one NIC NAT setup, is this right ?

> nat_router = 10.0.13.254 aft0:2

It this the default route on both real servers ?

>      address = 10.0.13.20 aft0:1
>      vip_nmask = 255.255.255.0
>      fwmark = 1
>      port = 80

This is inconsistant. You have to choose between VIP:port and fwmark.

>      load_monitor = none

There used to be a bug with that. Please check you have 2 nanny
processes once pulse is started. If none, replace load_monitor = none by
load_monitor = uptime (not a matter for the moment).


-- 
Sébastien Bonnet
  Centre de contacts - Experian France