> ADSM Development is looking into a design for centralized logging of client
> errors so they could be queried by administrators, and we would like to get
> some customer input to the design.  The initial requirement is for
> centralized logging pertaining to scheduled commands with potential to expand
> that to all ADSM client commands, whether scheduled or not.  We're interested
> in hearing what you would like to see reported from the clients to the server
> and what type of queries would be the most valuable to you.  We would also
> appreciate any other input you would like to provide to help make this design
> as crisp as possible.  Thank you very much.

Some thoughts on this:

As has been stated before, the ability to filter would be fine. The default
should be to report the error (non informational) messages.

The use would probably be similar to the "QUERY EVENT" command: one should be
able to concentrate on nodename(s), time/date range, exception-only messages,
... .

I am not sure, whether all messages are totally self-explanatory; one might
want to see it's context, e.g. messages like
"Incremental backup of file system: '/'"
indicating the start of a file system's incremental backup.

One might also want to know, what the initiating command looked like, which
options where in effect, which version/release/mod.level the client has.

-----------------------------------------------------------------------
Reinhard Mersch                     Westfaelische Wilhelms-Universitaet
Reinhard Mersch                     Westfaelische Wilhelms-Universitaet
Universitaetsrechenzentrum, Einsteinstrasse 60, 48149 Muenster, Germany
E-Mail: mersch AT uni-muenster DOT de                   Phone: +49(251)83-2488