>
> This listing leads me to believe your atalkd.conf file was/has
> changed. netatalk will do that when it restarts. Here's what mine
> looks like for a multiple NIC machine:
>
> cat /usr/local/etc/netatalk/atalkd.conf
>
> ...
> eth0 -seed -phase 2 -net 21-50 -addr 48.29 -zone "ethernet"
> eth1 -seed -phase 2 -net 51-60 -addr 51.44 -zone "ethernet"
> #eth2 -phase 2 -net 65280-65534 -addr 0.0
> eth2 -seed -phase 2 -net 61-70 -addr 61.44 -zone "ethernet"
>
> My guess is your file looks a lot like the commented eth2 device
> above. That is, you see something like:
>
> eth0 -phase 2 -net 65280-65534 -addr 0.0
> eth1 -phase 2 -net 65280-65534 -addr 3.0
>
> With multiple NICs you'll need to seed them with an address not in
> conflict with some other appletalk router. The safe bet, if a
> smaller network, is to choose a net and address that is outside the
> range of all of the other machines on the networks and then restart
> all of the machines to force them to get new appletalk addresses - or
> simply make sure all of the other machines are powered off when you
> restart the netatalk server. If there is a router conflict, the
> interface will not start its appletalk side of things. The IP side
> of netatalk is not affected by this in any way (AFAIK).
>
> Oh, I also set the permissions on /usr/local/etc/netatalk/atalkd.conf
> to 440 to prevent it from being overwritten at startup time. You
> might want to do this once you have it working again.
>
> One last observation, sometimes, when my server starts, the switch
> that it is connected to is not quite with it yet (it's still
> learning) when the server gets to the part of starting the netatalk
> server software. This will generally cause the appletalk side of
> things to fail just like your message above, but if I set the
> permissions on the file as shown above - it seems to force the
> appletalk side to accept the settings provided. This is just my
> theory - but as the switch has spanning tree enabled, I believe the
> solution seems appropriate.
>
> Regards,
>
> Joe Maus
>
Joe,
Thank you very much for your insightful info.
Here is what I had:
- eth0: connected to a network with no known AppleTalk router.
- eth1: connected to second network with an NT-based
AppleTalk router on the network. This network had net range
201-201 and 4 zones: "Zone 1", "Zone 2", "Zone 3", "Zone 4"
1a) I created this atalkd.conf file before running atalkd
because I wanted to use the default:
eth0
eth1
1b) file atalkd.conf about 30 sec after atalkd started:
eth0 -phase 2 -net 65280-65534 -addr 65280.243
eth1 -phase 2 -net 65280-65534 -addr 0.0
Note: eth1 did not get a net.node number and zone assigned
by the AppleTalk router!!!
2) file syslog.txt
<6> Jun 26 17:49:37 syslogd: restart
<5> Jun 26 17:49:37 syslogd: klogd startup succeeded
<6> Jun 26 17:49:38 kernel: Kernel logging (proc) started.
<6> Jun 26 17:49:45 atalkd[13018]: restart (1.5.3.1)
<6> Jun 26 17:49:46 atalkd[13018]: zip_getnetinfo for eth0
<6> Jun 26 17:50:05 last message repeated 2 times
<6> Jun 26 17:50:15 atalkd[13018]: as_timer multiple interfaces, no seed
<6> Jun 26 17:50:15 atalkd[13018]: as_timer can't configure eth0
<6> Jun 26 17:50:15 atalkd[13018]: as_timer waiting for router
<6> Jun 26 17:50:15 atalkd[13018]: ready 0/0/0
<3> Jun 26 17:51:09 afpd[13162]: Can't register QNTM610041:AFPServer@*
<6> Jun 26 17:51:09 afpd[13162]: ASIP started on 10.33.1.39:548(1) (1.5.3.1)
<6> Jun 26 17:51:09 afpd[13162]: uam: loading
(/etc/atalk/uams/uams_guest.so)
<6> Jun 26 17:51:09 afpd[13162]: uam: uams_guest.so loaded
<6> Jun 26 17:51:09 afpd[13162]: uam: loading
(/etc/atalk/uams/uams_clrtxt.so)
<6> Jun 26 17:51:09 afpd[13162]: uam: uams_clrtxt.so loaded
<6> Jun 26 17:51:09 afpd[13162]: uam: "Cleartxt Passwrd" available
<6> Jun 26 17:51:09 afpd[13162]: uam: "No User Authent" available
3) Ran ifconfig -a after running afpd
eth0 Link encap:Ethernet HWaddr 00:C0:B6:09:4E:F9
inet addr:10.33.1.39 Bcast:10.33.31.255 Mask:255.255.224.0
EtherTalk Phase 2 addr:65280/243
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:5560835 errors:429 dropped:0 overruns:0 frame:429
TX packets:168966 errors:0 dropped:0 overruns:0 carrier:0
collisions:15290
RX bytes:992969092 (946.9 Mb) TX bytes:56896651 (54.2 Mb)
eth1 Link encap:Ethernet HWaddr 00:03:47:9A:38:61
inet addr:192.168.48.190 Bcast:192.168.51.255 Mask:255.255.252.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:4193810 errors:139 dropped:0 overruns:0 frame:139
TX packets:68823 errors:0 dropped:0 overruns:0 carrier:0
collisions:264
RX bytes:632967734 (603.6 Mb) TX bytes:19982565 (19.0 Mb)
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
EtherTalk Phase 2 addr:0/0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:32713 errors:0 dropped:0 overruns:0 frame:0
TX packets:32713 errors:0 dropped:0 overruns:0 carrier:0
collisions:0
RX bytes:6165150 (5.8 Mb) TX bytes:6165150 (5.8 Mb)
----------------------------------------------------------
I tried seeding the two interfaces before, but this led to
serious consequences for my lab. The lab had many test machines,
and each one tried to configure itself as a router with distinct net
ranges. My netrange assignment algorithm was based on a modified version
of the program getzones. Each machine ran this program, which broadcast
a request to get netrange and zones from all routers on the network.
All routers on the network would report this info within 30 sec.
All responses that reached after 30 sec were ignored.
Based on this info, the machine chose a netrange that was not in
conflict with existing netranges that it received so far.
(See http://www.transit.hanse.de/netatalk/router.html)
Unfortunately, some routers in my lab did report after 30 seconds.
Hence, the netrange assignment eventually became conflicting,
causing lots of RTMP error messages, which finally brought the
network to a halt.
At last I decided to abandon this scheme and opted for a
default configuration (auto configuration).
Now my question is:
For multiple interfaces connected to the same network, can I
configure the interfaces with default settings (i.e., no seed)?
Thanks
Hoc
|