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

List:       linux-ha
Subject:    Re: [Linux-HA] inactive node answers arp requests for failover ip
From:       nils toedtmann <ntoedtmann () marcant ! net>
Date:       2003-07-30 12:50:25
[Download RAW message or body]

On Wed, Jul 30, 2003 at 04:29:32PM +0900, Horms wrote:
> On Tue, Jul 29, 2003 at 08:44:27AM -0600, Alan Robertson wrote:
> > Hi Nils,
> > 
> > nils toedtmann wrote:
> > >Hello,
> > >
> > >i've set up a ha firewall with heartbeat & freeswan. Works fine,
> > >but once in a while __both__ nodes answer the routers arp request
> > >for the failover ip. In case the arp-reply from the wrong, inactive node
> > >("node2") comes later, it wins and the ip packets go to the it's NIC. 
> > >Since the failover ip is not bound to node2's interfaces, the packets 
> > >get discarded. Absolutely nothing appears in the logfiles.
> > 
> > OK...
> > 
> > Any time heartbeat issues any commands that do anything to anything, it 
> > logs them.  So, if the behavior comes and goes without any log entries, 
> > heartbeat is probably not causing it.  I would suspect the set of services 
> > and patches you're currently running -- since I don't know of anyone else 
> > who has reported this problem.  Horms or lmb might be of more help on this 
> > one though....
> 
> A few things spring to mind: 
> 
> * Check interfaces: ip addr sh

As i said in my original mail, that's ok (inactive node not bound to
any VIP).
 

> * Check for static arp entries: arp -a

Where? On the router? Anyway, we don't use static arp entries, so if
heartbeat doesn't use them, there are none.


> * What kernel is being used? 
>   Sometimes strange kernels do strange things.

As i said in the initial posting:

> > The systems: redhat8, monolithic kernel (2.4.21 with several patches
> > like freeswan, lids, pax, some netfilter patches), 
> > heartbeat-1.0.3-1.rh.8.0.1
> > from <http://www.ultramonkey.org/download/heartbeat/1.0.3/redhat_8.0/>

Since the inactive node runs on a redhat errara kernel, the problem
disappeared. Now we try to figure out which kernelpatch is evil. 

(compile kernel, boot it, wait for two hours to see if arp-problem occurs.
 compile next kernel ...)
 

/nils.


--
nils toedtmann
department for technical paranoia
marcant internet-services gmbh <http://www.marcant.net/>
_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
[prev in list] [next in list] [prev in thread] [next in thread]