Comma-separated ADDRFAM value breaks resolvconf

[jaqxues]

We have the following configuration:

/etc/network/interfaces

# eth0 is slave
# eth1 is slave

auto bond0
iface bond0 inet manual
	bond-slaves eth0 eth1
	bond-mode 802.3ad
	bond-xmit-hash-policy layer3+4
	bond-miimon 100
	bond-lacp-rate 1
	bond-min-links 1
	post-up echo 1000 > /sys/class/net/${IFACE}/bonding/ad_actor_sys_prio

iface bond0 inet6 manual
	# Explicitly disabling acceptance of RAs can race, disable IPv6 on this interface instead.
	post-up echo 1 > /proc/sys/net/ipv6/conf/${IFACE}/disable_ipv6

# eth2 is slave
# eth3 is slave

auto bond1
iface bond1 inet manual
	bond-slaves eth2 eth3
	bond-mode 802.3ad
	bond-xmit-hash-policy layer3+4
	bond-miimon 100
	bond-lacp-rate 1
	bond-min-links 1
	post-up echo 1000 > /sys/class/net/${IFACE}/bonding/ad_actor_sys_prio

iface bond1 inet6 manual
	# Explicitly disabling acceptance of RAs can race, disable IPv6 on this interface instead.
	post-up echo 1 > /proc/sys/net/ipv6/conf/${IFACE}/disable_ipv6
# The loopback network interface
auto lo
iface lo inet loopback

auto vmbr0
iface vmbr0 inet static
	bridge-ports bond1
	bridge-stp off
	bridge-fd 0
	address 10.233.42.19/24

iface vmbr0 inet6 manual
	# Explicitly disabling acceptance of RAs can race, disable IPv6 on this interface instead.
	post-up echo 1 > /proc/sys/net/ipv6/conf/${IFACE}/disable_ipv6

auto vmbr1
iface vmbr1 inet static
	bridge-ports bond0
	bridge-stp off
	bridge-fd 0
	address xxx.xxx.xxx.xxx/26
	gateway xxx.xxx.xxx.1
	dns-nameservers xxx.xxx.xxx.xxx xxx.xxx.xxx.xxx
	dns-search vis.ethz.ch

iface vmbr1 inet6 static
	address xxxx:xx:xx:....::119/118
	# ifupdown2 enables forwarding automatically on bridges which have an IP assigned.
	# Set accept_ra to 2 for them being accepted with forwarding enabled.
	post-up echo 2 > /proc/sys/net/ipv6/conf/${IFACE}/accept_ra

For vmbr1, we specify the dns-nameservers etc. We include inet and inet6 configurations. This makes ifupdown2 pass an ADDRFAM value to resolvconf that looks like inet,inet6 which is not supported by the resolvconf hook for ifupdown2.

Is this an issue of ifupdown2 or resolvconf? I would argue that ifupdown2 should follow ifupdown in this case to not break compatibility.

[julienfortin]

@jaqxues what does ifupdown1 do in that case?
Maybe you can write your own wrapper around resolvconf to ignore it.

[jaqxues]

I tested what ifupdown1 does with bonds (not bridges, so slightly different network interfaces file). It calls the hook twice, once with ADDRFAM=inet and once for ADDRFAM=inet6.

Currently ifupdown2 seems to even ignore the IF_ADDRESS for inet6 as seen in these configurations...

/etc/network/interfaces.d/bond0

# eth0 is slave
# eth1 is slave


auto bond0
iface bond0 inet static
        slaves eth0 eth1
        bond_mode 802.3ad
        bond_xmit_hash_policy layer3+4
        bond_miimon 100
        bond_lacp-rate 1
        echo 1000 > /sys/class/net/bond0/bonding/ad_actor_sys_prio
	address   xxx.xxx.xxx.xxx
	network   xxx.xxx.xxx.xxx
	netmask   xxx.xxx.xxx.x
	broadcast xxx.xxx.xxx.xxx
	gateway xxx.xxx.xxx.1
	dns-nameservers xxx.xxx.xxx.xx xxx.xxx.xxx.xx
	dns-search vis.ethz.ch

iface bond0 inet6 static
	address xxxx:xx:xx:xx/118

[julienfortin]

So, if present, we have to run all hook for each address family? It would be better if each hook would handle the comma syntax though. But if this is what ifupdown1 does, then i'm okay with it, feel free to submit a PR.

[jaqxues]

Apparently resolvconf has a fix on debian that can handle comma separated...