[286] in Info-AFS_Redistribution
Cross cell monitoring survey.
daemon@ATHENA.MIT.EDU (Steve Mattson)
Fri Aug 16 14:39:29 1991
To: Info-AFS@transarc.com
Cc: afs_systems@caen.engin.umich.edu
Date: Fri, 16 Aug 91 13:37:58 -0400
From: Steve Mattson <hobbes@caen.engin.umich.edu>
Do other AFS sites out there do any kind of active monitoring
of external cells on a regular basis? Is it advantageous to
do anything beyond placing cells in /afs from Transarc's
exportable cells list? Occasionally we're interested in
things like number of fileservers other cells have, in a
generic, statistical way, and we go get that type of
information as needed only. Is anyone using tools to do this
with any regularity, and if so why? It would be nice to
have the information at times, but I wouldn't want to discourage
people from announcing their cells if as soon as they do so
something akin to the "n -> LARGE number of xloads monitoring
a main server for reliability" problem were to take place.
Would people consider this something that grand.central.org
might have a purpose in doing as a net.service?
Alternatively, what level of monitoring your cell from the
outside would you consider intrusive behavior? Would you
say that some of the AFS commands provide too much
information between cells without requiring authentication?
Would you consider any kind of outside mointoring not related
to use of the filesystem directly abusively taxing on your
cell's servers?
I'd like to hear from anyone doing this or really opposed
to it, and I can summarize for everyone if you email me.
Personally I'm not convinced either way, yet.
Thanks,
Steve Mattson
University of Michigan, College of Engineering
