In message <9411072224.AA16523@angst.microunity.com>, Eric Murray writes:
Eric> My question is: is anyone actively using netguard to gather stats?
Eric> Is anyone maintaing the code?
The netguard code is not very useful in its current state. Besides
some coding errors, there are some fundamental things that could be
handled much better. The current netguard implementation reads
configuration files that define which process is responsible to
collect data from which host. The mapping how these requests are
processed is done by the person who writes the configuration files.
A far more better scheme would just define requests for different
variables so that an `optimizer' would be able to create a good
`measurement plan'. The optimizer should make sure that nothing is
retrieved more than once and that get requests (e.g. SNMP requests)
retrieve as many variables in one message/reply interaction as
possible.
The current implementation is not a good starting point to implement
these ideas, as the tcl code really suffers from the lack of
structures in tcl (see the debate on scripting languages in you
favorite newsgroup.)
I think, a new implementation should be written in C and just use very
simple scotty scripts to aggregate raw data in interesting ways. As I
don't have time to do something like this, I just let the netguard
code in the scotty sources so that someone who is willing to spend
time on this could look at it.
On Mon, 07 Nov 1994 "John P. Rouillard" <rouilj@cs.umb.edu> said:
John> Also, is this the newest release of the code? I somehow got the
John> impresion that it developed a bit beyond what was in the release.
The last changes happend when I changed the arguments for the tkined
dialogs. I never touched the source since then.
John> One additional thing. The changes to the snmp stuff in scotty
John> 1.1.1 hasn't been propagated to netguard. I am starting on
John> some of that work, but it will be a while before I have
John> anything decent.
Replacing the old snmp stuff shouldn't be too hard based on
scotty-1.1.1, since you can replace the `snmp getbulk' with the new
walk session command. You will need some additional calls to lindex to
get the old behavior, but this should be quite obvious.
Juergen