[hobbit] Server-side extension scripts: shell vs. C programming

[Henrik Stoerner]

On Thu, Jun 28, 2007 at 02:12:33PM -0400, Gary Baluha wrote:
> I remember reading somewhere in the Hobbit documentation that when an
> extension script starts to do a lot of things, it should be coded in a
> compiled language such as C, instead of as a shell script.  I have a custom
> script that takes a lot of data and converts it into NCV graphs, and I
> believe it is at the point where I should consider rewriting it in C.

It really depends on how much work it does, and how often it runs: Do
you have 10 hosts calling into Hobbit with this kind of data, or 1000 ?
Your system load will be much less with a compiled C program than with
a shellscript or Perl program, but it should be considered against the
amount of work needed to re-implement your script.

> Before I get to far into it, is the hobbitd_sample.c something that I should
> look at for this?  I'm not sure if I'm reading the documentation on it
> correctly, if it is a good example for an external script.

hobbitd_sample shows how your program is fed data from Hobbit. I'm
not sure quite how your program currently works - does it run via
the "--extra-script" option to hobbitd_rrd ? In that case it might
be easier for you to just convert your current script into a C program
and still run it via the extra-script option.

If you do decide that the extra-script method is too heavy, then
modifying the hobbitd_rrd module is the easiest way to go. You'll have
to create a new routine to parse the data - start with a simple one 
like hobbitd/rrd/do_bbtest.c, which just finds a single value in the
"bbtest" status message, and updates and RRD file with it. Then you 
must add two lines to hobbitd/do_rrd.c: An "#include rrd/myfile.c" to
pick up the parser routine you just wrote, and a line near the bottom
of the file where you invoke your parser whenever the correct "status"
or "data" message arrives.


Regards,
Henrik