[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha
Subject:    RE: asymetric behaviour
From:       "Soffen, Matthew" <msoffen () iso-ne ! com>
Date:       2002-09-24 17:23:25
[Download RAW message or body]

What do your logs show ?

Does it take longer for 1 to realize that 2 is down ?

And what is that 15 seconds for (time to respond to pings or something else)
?
 
Matt

-----Original Message-----
From: Werner Schultheiss [mailto:ws@br-online.de]
Sent: Tuesday, September 24, 2002 1:16 PM
To: linux-ha@muc.de
Cc: Barbara Haeberlin
Subject: asymetric behaviour


Hi,

not a big problem, but wondering why:

I'm setting up two heartbeated firewalls with
2 identical systems (x330, Kernel 2.4.18, heartbeat
0.4.9.1-1) and am wondering about reproducable
asymetric time behaviour in nice_failback mode:

(a) node 1 goes down, node 2 takes over : 5 seconds
(b) node 2 goes down, node 1 takes over : 15 seconds

Monitoring that for several times showed that in (b) 
this time difference is while running the script ifstat
after ip_request on node 1 :
   /etc/ha.d/rc.d/ip_request ip_request
   /etc/ha.d/rc.d/ifstat ifstat
while in case (a) those scripts are not called resp. 
don't appear in /var/log/ha-log. 

Does anybody have any idea, why this happens ?

Thanks in advance,
Werner

-------------------
configuration below
-------------------

ha.cf
-----
serial /dev/ttyS0
watchdog /dev/watchdog
keepalive 1
deadtime 3
nice_failback on
baud 19200
debugfile /var/log/ha-debug
udp eth1
udpport 694
logfile /var/log/ha-log
logfacility local0
node fwsin-1
node fwsin-2

haresources
-----------
fwsin-1 10.65.3.244/24  dmz ipfw
-- 
.....................................................
Werner Schultheiß              fon: +49 89 5900-16031
BR HA Multimedia               fax: +49 89 5900-16000
Rundfunkplatz 1                email: ws@br-online.de
D-80300 München
.....................................................
[prev in list] [next in list] [prev in thread] [next in thread]