On 10/09/2021 00:38, mike tancsa wrote:
On our standard nanobsd router image we use, I noticed failover routing
broke all of a sudden. We are not using multiple FIBs, but somehow we
now have two default routes I am guessing because of multipath. (Note,
we just started testing RELENG_13 so its possible I am making bad
assumptions somewhere.)
default 10.255.255.1 UGS 6 1500 igb0
default 10.1.0.1 UGS 17 1500 tun10
I am pretty sure from the code base from 3+ days ago, I would not see
this condition.
I just want to add that I've been seeing the same behavior on CURRENT for a few
months now.
I think that ppp probably needs some sort of an update to be aware of multipath.
BTW, net@ may be a good mailing list to discuss this as well.
In the past, my nano image would get a default route via DHCP. If that
path was not functional or blocked, my monitoring software would detect
the failure and fire up PPP which has
add! default HISADDR
That now seems to add a second default route instead of getting rid of
the old one.
if I add net.route.multipath=0
I get back the old behaviour
According to the ppp docs
If the add! command is used (note the trailing "!"), then if the
route already exists, it will be updated as with the `route change'
command (see route(8) for further details).
Where as now, it doesnt seem to replace it, it seems to just add another
default route.
For my purposes, setting net.route.multipath=0 seems to work just fine,
but I wonder if this warrants a note in ppp's docs since it no longer
seems to work that way in RELENG_13 ?
--
Andriy Gapon
