<snip>
I have a couple of thoughts on things raised in this thread:
1. When I first became a *NIX admin I went from being an assembly
language programmer of code for embedded systems to administrating
a system running 4.1a BSD. What saved my bacon was the KWIC index
(KeyWord In Context) for the BSD man pages and documents.
I believe that some sort of comprehensive index to the Linux man
pages and the LDP generated docs would allow many people to resolve
their own issues. I do not know if ptx(1) would meet this need, but
something like a KWIC index for the Linux documents would be of huge
assistance.
2. In a past job I had occasion to give a bit of a class to new system
administrators titled something like "Host Based Diagnosis of
Network Problems." The part of the information which seemed to be
most helpful was telling these admins was for any problem reported
to them was to insist on details indicating a fault . This included
a precise description of what the user was doing when the fault
occurred, the precise text of any error messages received, and any
log entries which resulted. There was some complaints that the
users would know into which log files to look.
On that point all I could say was a computer is just a tool and
people should know the rudiments of using their tools and that
includes documentation, where the log files are and generally what
type of logs go in which log files. Syslog.conf is a key resource
to figure that out. Now "mere users" can't read the logs, usually,
but on this list people can read logs for the most part.
It occurs to me that people who ask for help should do these things
at a minimum. Precision is important, it is a great assistance to
those of us who might help. It is key that we who try to help
should not have to spend a lot of time trying to figure out what
question is really being asked. Usually there multiple possible
causes, so multiple possible responses. Make it easy on us, and you
will get quicker and more accurate answer to questions.
dlg
