On 2008-10-14, Alan McKinnon <[EMAIL PROTECTED]> wrote:
> On Tuesday 14 October 2008 15:59:54 Grant Edwards wrote:
>
>> > from 'cat /etc/rc.conf'
>> >
>> > # Some people want a finer grain over hotplug/coldplug. rc_plug_services
>> > is a # list of services that are matched in order, either allowing or
>> > not. By # default we allow services through as rc_coldplug/rc_hotplug has
>> > to be YES # anyway.
>> > # Example - rc_plug_services="net.wlan !net.*"
>> > # This allows net.wlan and any service not matching net.* to be plugged.
>> > rc_plug_services=""
>>
>> 1) None of that text appears in my /etc/rc.conf
>
> Oops, slipped my mind - that text shows up in openrc and baselayout-2 which
> you might not be using yet
>
>> 2) I don't understand how the above applies.
>>
>> 3) Has that changed recently thus breaking support for bridge
>> interfaces?
>
> Inferred from the error message - if you prevent any hotplugging from
> occurring for a virtual interface, the problem ought to be solved.
I want hotplugging to work for br0.
> I'm not sure about recent thinks breaking the bridge, my only
> recent experience is that bridges tend to break themselves in
> frustrating ways, ably helped along by virtualization software
> <bang head><bang head>
It worked the last time I restarted br0, so I'm going to leave
it alone for a while. It turns out it doesn't always fail with
that message either. Sometimes it fails after dhcpcd fails
like this:
Oct 14 08:41:10 grante br0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM
feature.
Oct 14 08:41:10 grante device eth0 entered promiscuous mode
Oct 14 08:41:10 grante br0: port 1(eth0) entering learning state
Oct 14 08:41:10 grante rc-scripts: ERROR: net.br0 is already starting.
Oct 14 08:41:11 grante dhcpcd[15052]: br0: dhcpcd 4.0.2 starting
Oct 14 08:41:11 grante dhcpcd[15052]: br0: broadcasting for a lease
Oct 14 08:41:25 grante br0: topology change detected, propagating
Oct 14 08:41:25 grante br0: port 1(eth0) entering forwarding state
Oct 14 08:41:38 grante dhcpcd[15052]: br0: offered 10.1.1.19 from 10.1.1.1
Oct 14 08:41:38 grante dhcpcd[15052]: br0: checking 10.1.1.19 is available on
attached network$
Oct 14 08:41:41 grante dhcpcd[15052]: br0: timed out
or sometimes like this:
Oct 14 08:43:08 grante br0: port 1(eth0) entering learning state
Oct 14 08:43:08 grante br0: port 1(eth0) entering disabled state
Oct 14 08:43:08 grante device eth0 left promiscuous mode
Oct 14 08:43:08 grante br0: port 1(eth0) entering disabled state
Oct 14 08:43:08 grante br0: Dropping NETIF_F_UFO since no NETIF_F_HW_CSUM
feature.
Oct 14 08:43:08 grante device eth0 entered promiscuous mode
Oct 14 08:43:08 grante br0: port 1(eth0) entering learning state
Oct 14 08:43:08 grante dhcpcd[15696]: br0: dhcpcd 4.0.2 starting
Oct 14 08:43:08 grante dhcpcd[15696]: br0: broadcasting for a lease
Oct 14 08:43:10 grante dhcpcd[15696]: br0: received SIGTERM, stopping
Oct 14 08:43:10 grante br0: port 1(eth0) entering disabled state
Oct 14 08:43:11 grante device eth0 left promiscuous mode
Oct 14 08:43:11 grante br0: port 1(eth0) entering disabled state
Oct 14 08:43:11 grante rc-scripts: We only hotplug for ethernet interfaces
It's probably dhcpcd's fault -- several times in the past it
has broken after an update because the default behavior has
been changed in a way that's incompatible with previous
versions.
--
Grant Edwards grante Yow! I'm using my X-RAY
at VISION to obtain a rare
visi.com glimpse of the INNER
WORKINGS of this POTATO!!
