I have a private ethernet LAN and Cayman 3220H DSL modem/router/hub. When idle the computers run Setiathome under the control of SetiQueue on a Win2K Pro machine. SQ keeps track of things using netbios names, if it can get them, or IP addresses.

Early this morning, SQ began replacing the usual names/IPs with names of the form hostx-null.null.bellsouth.net as each machine transmitted a completed Seti workunit. The "x" corresponds to 192.168.1.x and Bellsouth is my DSL ISP. It is now beginning revert back to the original names/IPs as machines with the weird names complete the first workunit under that name. Some machines have more than one IP (more than one NIC) and several are dualies (one of those hyperthreads). All instances of Seti@Home on a given machine switch names at once. The weird names do not appear in the Cayman DCHP client list and are not pingable. I continue to be able to ping machines by the names I assigned or by IP if samba isn't fully operational (still haven't redone on the linux machines after installing new distros/versions).

Has anyone seen anything like this? Is this some smb/nmbd weirdness or, perhaps something sinister?

The appearance of the weird names (they are not netbios names, after all) seems to have something to do with the RH8 netconfig tool settings. Things are better if I select DHCP but do not specify a hostname and uncheck "automatically obtain DNS information from provider." I had thought the latter would read the name table on the DHCP server - NOT. To be able to see the Internet, I have to specify my router address as the "primary DNS server." This appears to put a null domain name in the DNS search path and is contrary to the directions provided by RH.

Not specifying the optional DHCP hostname on the Macs also seems to prevent their IP addresses from being replaced with what I had called weird netbios names - what ever they are, they aren't netbios names.

This is expected behaviour...

Um, that is.. I didn't expect it myself, until I read through the dhcp scripts, and discovered that...

If you don't set the hostname the script runs dhcpcd -H, where -H means 'set the hostname of this machine to the hostname supplied by the dhcp server.'

You can either - supply a hostname, edit /etc/sysconfig/network-scripts/ifup and search for :

if need_hostname; then
PUMPARGS="${PUMPARGS} --lookup-hostname"
DHCPCDARGS="${DHCPCDARGS} -H"
fi

and remove the -H

:P

Slick.

Slick W,

I couldn't tell any difference after making the change you suggested but have started poking around in the scripts.

I am still confused about the different names that can be assigned to a host or NIC and where you do it for the different OSes (Linux, OSX, OS/2, Win2K) and how to make sure everything gets updated with changes are made.

I'll spare everyone a dump of the details but would appreciate an explanation of the differences between the "hostname", DHCP optional "hostname", and the NetBios name. BTW, none of these are the weird hostnames that sometimes show up as hostnames when logging on a linux or OSX machine and in the SetiQueue client lists.

R.