dhcpcd-discuss

> Sounds like ral0 is reporting link up/down events.
> On NetBSD, you could see this with the route(8) monitor command. syslog
> messages are normally generated as well.

I didn't see any syslog messages.

"route -n monitor" is quiet, but I will watch for awhile.

> A good way to troubleshoot dhcpcd issues is simply to add the word debug
> on a line by itself in /etc/dhcpcd.conf (or /usr/pkg/etc/dhcpcd.conf)
> and then check syslog for extra messages. You can also keep dhcpcd in
> the foreground and specify debugging manually like so:
> dhcpcd -dB

Thanks. I ended up using -dB.

> Whilst your issue is isn't a dhcpcd issue itself, you can make dhcpcd
> ignore the link up/down events by adding nolink to dhcpcd.conf or giving
> the -K command line option. This makes dhcpcd behave more like dhclient
> which has no link carrier support.

Thanks for the reminder. I now see I use --nolink in my dhcpcd_flags in 
my rc.conf on a different system. My comment there is simply:
# --nolink used because bad device shows "no network" sometimes

That works for me here too.

(I also have --noipv4ll in that configuration. I don't recall the 
details, but I recall it was assigning some zeroconf temporary address 
and was failing to assign real address.)

> Of course the longer term fix would be to fix ral(4) in NetBSD. I say 
> this because I have both ral(4) and ath(4) in the same box and ral(4) 
> behaves exactly as you describe.

Okay, I may later file a PR for that.

Thank you very much.