On Thu, 23 Mar 2006, Andreas Lundqvist wrote:
It still doesn't on an physical interface thats already up
but on loopback it never arp's regardless of the -arp/arp
On linux the kernel doesn't care which interface has the IP,
it just replies if the machine has the IP. In HPUX the IP
must be tied to the device.
I did a tcpdump on the interface when I took up the
loopback interface alias and there wasn't any arp
request/broadcast on the interface that carries the
did you try these?
Does it matter what netmask/broadcast I use on the loopback alias?
I've seen examples that looks like this:
vip: 10.10.0.10 netmask 255.255.255.0 broadcast 10.10.0.255
only for the VIP on the director for LVS-NAT
On realserver it the looks like this:
10.10.0.10 netmask 255.255.255.255 broadcast 10.10.0.10
for the VIP on all machines in LVS-DR
But an 255.255.255.255 netmask on HPUX loopback gets an error message so I'm
forced to use my normal netmask.
there is supposed to be someway to handle this, but I don't
remember what you do to trick the machine.
Do you see any problems with this?
not if it's working.
if you're still getting load balancing after some testing (a
week, month?), could you send us a description of your setup
o ipvsadm on the director
o number of NICs/networks on director
o how you setup the VIP on the realservers.
o what you did to convince yourself that you'd solved the
arp problem (other than it works - which you can do if
you're lucky without solving the arp problem explicitely)
I'll put it in the HOWTO. HPUX hasn't been used much on
Joseph Mack NA3T EME(B,D), FM05lw North Carolina
jmack (at) wm7d (dot) net - azimuthal equidistant map
generator at http://www.wm7d.net/azproj.shtml
Homepage http://www.austintek.com/ It's GNU/Linux!