[Linux-HA] Newbie: linux-ha R2 firewall ERROR:glib: Unable to send bcast [-1] packet(len=180): No such device

[Rois Cannon]

I have 2 Mandriva 2007.1 installed with 2 network cards each.
I'm using rpms:
libheartbeat-pils1-2.0.7-10mdv2007.1
libheartbeat1-2.0.7-10mdv2007.1
heartbeat-pils-2.0.7-10mdv2007.1
heartbeat-2.0.7-10mdv2007.1
libheartbeat-apphb0-2.0.7-10mdv2007.1
libheartbeat-stonith1-2.0.7-10mdv2007.1
heartbeat-stonith-2.0.7-10mdv2007.1

On top of a minimal install.

My goal is a 4 nic system: 2 ISP, 1 Lan and 1 DMZ on each manchine
(active/passive).

svr3:
eth0 192.168.151.3
eth1 192.168.0.11

svr4:
eth0 192.168.151.4
eth1 192.168.0.12



Currently ha.cf:
-----------------------------------------------------
keepalive 1
deadtime 3
hopfudge 1
udpport 1001
udp     eth0
debugfile /var/log/ha-debug
logfile /var/log/ha-log
logfacility     local0
node svr3.lan.ccccxc.net
node svr4.lan.ccccxc.net
-----------------------------------------------------

haresource
-----------------------------------------------------
svr3.lan.ccccxc.net 192.168.151.5/24/eth0
-----------------------------------------------------

Pretty minimal.  (Small bites since I'm new to this.)

If I unplug the cable to svr3 on eth0, svr4 takes over just fine
HOWEVER . . .  svr3 ha-debug log is repeating this message
------------------------------------------------------------------------
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: glib: Unable to send bcast
[-1] packet(len=180): No such device
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG: Dumping message with 12
fields
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[0] : [t=status]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[1] : [st=active]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[2] : [dt=bb8]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[3] : [protocol=1]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[4] :
[src=svr3.lan.ccccxc.net]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[5] :
[(1)srcuuid=0x80fe930(36 27)]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[6] : [seq=a7]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[7] : [hg=6]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[8] : [ts=4667327c]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[9] : [ld=0.02 0.04 0.01
2/69 9791]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[10] : [ttl=3]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: MSG[11] : [auth=1 bc792c68]
heartbeat[9017]: 2007/06/06_15:17:32 ERROR: write failure on bcast
eth0.: No such device
heartbeat[9017]: 2007/06/06_15:17:33 ERROR: glib: Unable to send bcast
[-1] packet(len=180): No such device
. . . 
------------------------------------------------------------------------

and syslog shows this:
------------------------------------------------------------------------
Jun  6 15:17:26 svr3 ifplugd(eth0)[2271]: Link beat lost.
Jun  6 15:17:32 svr3 ifplugd(eth0)[2271]: Executing
'/etc/ifplugd/ifplugd.action eth0 down'.
Jun  6 15:17:34 svr3 kernel: ACPI: PCI interrupt for device 0000:00:0d.0
disabled
Jun  6 15:17:34 svr3 ifplugd(eth0)[2271]: Program executed successfully.
Jun  6 15:17:35 svr3 kernel: PM: Writing back config space on device
0000:00:0d.0 at offset f (was 50a0100, writing 50a010b)
Jun  6 15:17:35 svr3 kernel: PM: Writing back config space on device
0000:00:0d.0 at offset 5 (was 0, writing de800000)
Jun  6 15:17:35 svr3 kernel: PM: Writing back config space on device
0000:00:0d.0 at offset 4 (was 1, writing b801)
Jun  6 15:17:35 svr3 kernel: PM: Writing back config space on device
0000:00:0d.0 at offset 3 (was 0, writing 2010)
Jun  6 15:17:35 svr3 kernel: PM: Writing back config space on device
0000:00:0d.0 at offset 1 (was 2100000, writing 2100017)
Jun  6 15:17:35 svr3 kernel: ADDRCONF(NETDEV_UP): eth0: link is not
ready
Jun  6 15:18:04 svr3 kernel: ADDRCONF(NETDEV_CHANGE): eth0: link becomes
ready
Jun  6 15:18:05 svr3 ifplugd(eth0)[2271]: Link beat detected.
Jun  6 15:18:06 svr3 ifplugd(eth0)[2271]: Executing
'/etc/ifplugd/ifplugd.action eth0 up'.
Jun  6 15:18:08 svr3 ifplugd(eth0)[2271]: Program executed successfully.
------------------------------------------------------------------------

It appears that ifplugd is doing ifdown on eth0 and heartbeat doesn't
like it.  I'm assuming I've done something wrong and it's a problem when
I try to udp and vip on both sides of the firewall (lan/wan):
Currently ha.cf:
-----------------------------------------------------
keepalive 1
deadtime 3
hopfudge 1
udpport 1001
udp     eth0 eth1
debugfile /var/log/ha-debug
logfile /var/log/ha-log
logfacility     local0
node svr3.lan.ccccxc.net
node svr4.lan.ccccxc.net
-----------------------------------------------------

haresource
-----------------------------------------------------
svr3.lan.ccccxc.net 192.168.151.5/24/eth0 192.168.0.10/24/eth1
-----------------------------------------------------

When I unplug either  nic (eth0 or eth1) on svr3 I want the failover to
take over and then go back to the master when it's fixed.  When I'm
udp'ing on both nic's the above error starts going crazy and svr4 does
NOT take over.  Presumable because one of the nic's is still talking.

I don't know anything about xml so I'm hoping I can use haresource
instead of the cib stuff.  Especially since it's just 2 nodes.

I've also tried to get ipfail and pingd to work and I'm probably just
not understanding how to make them do something.

Perhaps I should try to find the gui.  Doesn't appear to be in the MDK
rpm source.

A friend of mine has something similar running under R1 and said it
works good but I'd rather use R2 since it's more current.

Any suggestions would be greatly appreciated.
_________________
Rois Cannon 
http://mymissionchurch.org - Your life is going somewhere whether you
have a strategy or not. Experience the High Life.