Subject: Re: SUP divergences between netbsd and slave servers
To: George Michaelson <>
From: Luke Mewburn <>
List: current
Date: 01/16/1995 11:58:44
> I just did a sup -Nvo off It requested several major
> removals of stuff the same sup options re-request from the AU hosted
> sup server. I went with but I'm deeply confused.
> I don't understand how this happens. Is it possible the RMIT sup database
> is corrupted? Here's the list of stuff its asking for, the crypt files are
> ok, thats a non US sourced lib/libcrypt replacement I know I need.

If the sup mirror off fails for some reason, it usually
has to wait until the next day for it to be restarted. I restarted
the sup mirror by hand about two hours ago because I noticed that it
wasn't fully upto date. If you still have problems tomorrow (950107),
email me and I'll nuke the sup status files and start a fresh update.

Note for core team: for some reason I noticed that the entire netbsd-1.0
mirror was retreived. Were the status files on sun-lamp for these
corrupted? (I only ask because it's 80MB of stuff that is being
re-mirrored for no other apparent reason.)

> Does sup use anything apart from file stat timestamps (eg a simple checksum
> or hash function across the file) to check what needs updating with -o?

Just timestamps in supfilesrv database vs local sup database I think.

> I know, take it up with the RMIT people. But I still think the central sup
> repositary should know if the slaves are not in agreement...

(Ex-RMIT people. I used to work for RMIT Computer Science, and I'm
still a part time undergrad there. The machine is @RMIT because one of
my friends still there agreed to keep the machine under his desk)


Luke Mewburn                          UNIX Technical Support
<>              CPR Project, ITG
Phone: +61 3 634 2112                   Telecom Australia.