On 08.06.2011 22:31, Andrew O. Shadoura wrote:
Hello Andrew,
Package: ifupdown
Version: 0.7~alpha4
Severity: normal
Tags: experimental
ifdown executes 'ip addr flush dev $DEV', but does not set the link
layer to down using "ip link set eth0 down". This flushes the IPv6
link-local address from the interface, which essentially disables all
IPv6 on it until you manually add the link local address again. It
will even disappear from the net.ipv6.conf.$DEV tree, which means the
next ifup with a static inet6-stanza will fail with
error: "net.ipv6.conf.eth0.accept_ra" is an unknown key
Failed to bring up eth0.
Please execute "ip link set $DEV down" on ifdown.
Mmmm, but it does:
down
ip -6 addr flush dev %iface% scope global
ip link set dev %iface% down
Can you provide some more details?
You are right, it is another problem. The /etc/network/interfaces looks
like this
auto eth0
iface eth0 inet static
address 80.244.243.68
gateway 80.244.243.65
netmask 255.255.255.240
when I issue ifdown all addresses are flushed, then it tries to delete
the default route. Which is already gone since the flushing.
Configuring interface eth0=eth0 (inet)
run-parts --verbose /etc/network/if-down.d
run-parts: executing /etc/network/if-down.d/clamav-freshclam-ifupdown
run-parts: executing /etc/network/if-down.d/postfix
ip addr flush dev eth0
ip route del default via 80.244.243.65 dev eth0
RTNETLINK answers: No such process
rc!=0, execution stops. Moving the gateway to "up ip route add default"
to avoid it being deconfigured indeed shuts the interface down correctly.
But part of the bug still remains, the "ip addr flush" in the IPv4 part
flushes all IPv6 (inluding link-local) as well. Maybe you should change
that to "ip -4 addr flush" to be consistent.
Bernhard
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
