At 14:49 Uhr -0700 08.10.1996, Julian Elischer wrote:
>Stefan Bethke wrote:
>> netatalk-1.3.3 compiles, but atalkd isn't able to run (haven't looked
>> further why). netatalk 1.4b1 doesn't compile on 2.1.5-R (errors in
>> compiling afpd, something with getquota()).
>(just comment out all the quota stuff :)
>the patch in control.c in atalkd in 1.4 was based on the sam epatch on
>and allowed atalkd to run by default..
>it does run
>if you supply a conf file, even without the patch
>as that code isn't run in that case.
I supplied a conf file ("ed0 -phase 2"), but atalkd didn't get the zone
list from our seed router.
Most interestingly, the original patches applied to 2.1.0-R work on our
production system, which I wanted to up to 2.1.5-R; I think I wait a little
with that :-(
>I assume you have the FreeBSD patches for 1.3.3?
>the kernel code we have came from the 2.1 port of 1.3.3
>I ported it to 2.2 and checked it in..
I'll look at it when I have more time. The kernel patches for 2.1.0 are
definitly buggy (as far as I can tell), 'cause they have no means of
receiving or sending packets from/to a machine in the startup range (see
Inside AppleTalk, pp.4-8, "Node address aquisition on extended networks").
Also (maybe because of that), atalkd doesn't work on an extended network
(ethernet) without a seed router, that is, a network running constantly in
the startup range.
Currently, you need a seed router to use netatalk; running netatalk as a
seed router does not suffice, because Macs try to get network information
from the seed router, sending a broadcast GetNetInfo with source address in
the startup range (or the last known network address, which also doesn't
work with standard 4.4 routing). This seems to be true for at least Linux
(Please note the cross-post to netatalk-admins, hopefully we two are not
the only ones interdested in getting netatalk to work completely on
Stefan Bethke <email@example.com>