On Tue, Mar 28, 2000 at 12:04:33PM +0100, FSG STUDENT4 Ex Directory wrote:
> Unfortunately it doesn't do any replication, as I noticed
> today. _ALL_ three volumes are only stored on my non-SCM
> machine sf2. Exploring the location of a file on any of
> these volumes with "cfs whereis" shows the files always 
> being stored on sf2. The replicated volume is even replicated
> on sf2, so I have it twice on sf2. No data is actually 
> stored on the SCM (sf3), although the createvol_rep command
> shows: "Creating volume coda:root.0 on sf3".
> 
> Any ideas?

Huh? it sounds like your DNS lookup for `sf3' gave the ip-address of
server sf2.

In any case, did you run the createvol_rep scripts on the scm? You might
be able to replace the VSGDB entries with ip-addresses of the server to
avoid broken DNS lookups, I'm not sure if that works but I guess it
beats debugging your DNS setup.

Purging these bad volumes might be difficult, you'll probably have to
use `volutil purge <volname> <volid>' on sf2 and strip the related lines
out of the VRList.

> BTW, can I use any value for the identifiers in the VSGDB,
> or is E0000100 always non-replicated and E0000104 always
> replicated? Could I also use 1234567 als VSG addess?

No relation, these VSGDB addresses actually originate from the era that
Coda still supported multicast ip, E0000104 is 224.0.1.4. The code still
seems to be there, but it is probably broken.

Jan