Pulse problems

To: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: Pulse problems
From: "David D.W. Downey" <qixo@xxxxxxxxxxxxxx>
Date: Sun, 25 Jun 2000 00:06:46 -0500 (CDT)
::exasperated sigh:: Well.. I'm at my wits' end on this.

OK time to kick out to the group what's going on..

I was having troubles with 2.2.14 and lvs staying active. (network
timeouts up the ying yang!)

Now, I've upgraded to kernel 2.2.16-4, ipvsadm 1.1-4, piranha 0.4.16-2 and
now the problem is that pulse refuses to start. I'm getting
pulse: SIOCGIFADDR cannot assign requested address errors. (Bear in mind
that I have absolutely NO clue what any errors to expect when installing
and using LVS as this is my FIRST LVS ever!)

Now, from what I can understand in the kernel source it usually means that
it can not assign the PA address. If I am correct this means the
Permanently Assigned address. (IP?)

OK, so going on that surmise, my question becomes one of why is pulse
trying to assign the IPs when I already have the IPs up and running?

(Due to NDAs I can not release the actual IP(s) I am working with net-side
just yet, so I am using a few fakes.)

Here is what I have set up.

my primary node is 234.456.789  on eth0
my lvs IP is 234.567.799 on eth0:1
my internal IP is on eth1
and the NAT device is on eth1:1

I have the same thing configured on the backup node (234.567.778 eth0)

Now, I also have set up to rr over ssh (using openssh) to a web
server farm. I have ipchains configured and ipvsadm configured correctly.

Now bear in mind that this same exact set up works fine under 2.2.14 with
the exception of the network timeouts and wierd ass IP reassignments. (LVS
node 0 seems to drop out and LVS node 1 takes over, then they flip flop
type thing even though they are both running. Told you it was wierd.)

I tried ifconfig ethXX down all the interfaces and then run
/etc/rc.d/init.d/pulse start but it dies with the same message as before

This has become a MAJOR!!! concern now because I MUST have this up and
running correctly by no later than June 29th as I am leaving for North
Carolina and the company NEEDS to have this running BEFORE I can go.

Does anyone here have any clue what could be the problem? I am willing to
call and talk via cell phone (USA list members only. Sorry but I have free
Long Distance in the USA only, nothing personal.)

I am on Pacific time (5pm EST == 2pm PST).
Any takers?

David D.W. Downey              RHCE, UNIX/Linux Administrator
QIXO, Inc.                     Internet Security Specialist            (408) 514-6400

<Prev in Thread] Current Thread [Next in Thread>