carp regression in 9.1 ?

Fleuriot Damien ml at my.gd
Thu Mar 14 14:47:19 UTC 2013


Hello list,


I'm experiencing this odd behavior with 9.1 r24791 for amd64.


Find below exerpts from my rc.conf related to carp:

===
# production
ifconfig_carp22="inet 192.168.22.254/32 vhid 122 pass vrrp22 advskew 100"
ifconfig_carp32="inet 192.168.32.254/32 vhid 132 pass vrrp32 advskew 100"
cloned_interfaces="carp0 carp1 carp22 carp32"

# test interfaces, use different syntaxes
ifconfig_carp1="vhid 124 advskew 100 192.168.22.240 netmask 255.255.255.0"
ifconfig_carp0="vhid 140 advskew 50 pass dsds 192.168.22.250"
ipv4_addrs_carp0="192.168.22.241-249/24"
===


With all of these configuration lines, carp interfaces come up at boot.


The problem I experience is when destroying the carp interfaces and recreating them:
root at pf1-drt:~ # ifconfig carp1 destroy
root at pf1-drt:~ # ifconfig carp1 create
root at pf1-drt:~ # ifconfig carp1 
carp1: flags=8<LOOPBACK> metric 0 mtu 1500
	nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>


FreeBSD 8.3 doesn't exhibit the same behaviour, when destroying then recreating the interfaces, they are initialized correctly.


Additionally, note the syntax I used for my test interface carp0:
ipv4_addrs_carp0="192.168.22.241-249/24"

And the results at boot:
carp0: flags=49<UP,LOOPBACK,RUNNING> metric 0 mtu 1500
	inet 192.168.22.250 netmask 0xffffff00 
	inet 192.168.22.241 netmask 0xffffff00 
	inet 192.168.22.242 netmask 0xffffffff 
	inet 192.168.22.243 netmask 0xffffffff 
	inet 192.168.22.244 netmask 0xffffffff 
	inet 192.168.22.245 netmask 0xffffffff 
	inet 192.168.22.246 netmask 0xffffffff 
	inet 192.168.22.247 netmask 0xffffffff 
	inet 192.168.22.248 netmask 0xffffffff 
	inet 192.168.22.249 netmask 0xffffffff



I'm really confused here, is this intended behaviour ?

FreeBSD 8.3 certainly doesn't act this way.


Does anyone have an idea what's happening there ?



More information about the freebsd-net mailing list