Ive
sent in a request to division supervisors and should hear back from them by
Monday. Im happy to share the code (and receive feedback, its far from
tight software), the initial archiver (C code) simply uses OVsnmpRecv, parses
and loops through the varbinds. I did have to bastardize snmpprint.h to
make output is a desired format, but most of the guts was taken from the
/usr/OV/prg_samples techniques. The desired format is a well-defined
delimited ASCII text file which the PERL post processor will use as a seed to
create the command file (ASCII format) which can also be altered as
needed. Im not sure how NT users will fair trying to use my application
suite as it was built on UNIX and uses the 'snmptrap' stack for replaying the
traps ... YMMV.
I
could have coupled the post processor into the archiver, but keeping them
seperate allows for other functionality to be built on top of the
archiver. A good example of this would be something Fawad was looking
for, like a simple Tk or Java front end to view and scroll through an entire
data set which is only bounded by the amount of disk space you have to store
the trap data. We do use datawharehousing methodologies here which is
basically what we are talking about, but I always find them to be cumbersome
and tightly wound like most COTS solutions so writing something with a small
footprint is not only cheaper but more adaptable IMHO.
Jason Allison
Principal Engineer
ARINC
Incorporated