Am 21.01.2011 00:38, schrieb Andre Keller:
I have a strange problem with network connectivity on a device of mine.
The setup is carp on vlan on vr(4).
The problem is that the link runs for 10minutes, 10hours or 10days and
suddenly it stops working. Doing a ifconfig vr0 down ; ifconfig vr0 up
solves the problem temporarly but as you can imagine I'd like to have a
more permanent solution.
The problem is there are no obvious indications. The port on the switch
(C 2960) stays up (and it is not errdisabled), there are no errors.
Configuring the interface 100 full or autoselect (on both switch and
device) does not make a difference. The error counters on the switch as
well as netstat -i do not show any errors.
I setup the same configuration (carp on vlan on physical interface)
using em(4) and did not run into the problem yet (3 weeks up). So I
guess I could have something to do with vr(4).
The problem appeared first after updateing 4.8 to 20101222 snapshot, and
is still present with snapshot from this week. But prior 20101222
snapshot there were no carp and vlan interfaces, just an ip on the
physical interface. So I don't know if the problem is my configuration
or something that has changed in the code...
There was a well known bug that would cause vr devices to stop receiving
and sending traffic if a short cable was used to connect to the switch
or other device (like a DSL modem). I really don't know if this was
fixed at some time, but I think I still saw this bug on my router
running 4.8-stable (a soekris 5501). I dealt with it monitoring the
pppoe0 interface and doing the ifconfig up / down of the vr interface
used by pppoe0 via script as soon as the pppoe connection is lost.
As I worked around the problem I can't offer you a solution. Perhaps
this information gives a hint to someone who can solve this.
Kind regards
Markus
