Cacti and Orca

[cleikis]

Hi,

I'm looking for some assistance in integrating Orca and Cati for a performance reporting project I'm working on. I'm down to the last few pieces that need to fall in place (at least as I see it).

What I've done so far:

1) Oracallator.se is running on each Solaris host collecting data via the SE toolkit and storing data on a NAS mount.

2) The Orca daemon is processing the data and creating/updating the rrd files under the directory structure /orcadata/rrd/o_hostname/xxx.rrd

3) I'm running Cacti 0.8.6i and I've configured data templates for each Orca data source, graph templates, and an Orca host template.

When I add a new Orca host, Cacti will automatically create all of the graphs and data sources. The only problem is that it doesn't populate the path to the rrd file. It either tries to create <rra_path>/xxx.rrd or just leaves the field blank (though I'm not sure how it decides which to do because it never actually creates the rrd file). Everything works fine when I manually edit each data source that's created and put in the proper path to the right rrd file.

What I'm looking to do is:

1) Write a script to periodically scan the Orca data directory structure, look for new hosts, and add them to Cacti. I think this is possible using cron and the add_device.php script I found in these forums.

2) Write a script or plug-in to "fix" any paths to Orca RRD files based on the data template. Since I skipped any per-disk device and per-inteface RRD files, each template has a specific rrd file. (E.g. "Orca - Scanrate" data template should have the path /orcadata/rrd/o_hostname/gauge_scanrate.rrd)

I have no problem writing the scripts, but I'm had trouble deciphering which fields and tables have the information I need to gather to update the paths. I also need some advice on the best way to implement which hopefully doesn't require customizing Cacti code, making future upgrades more difficult.

I'm planning to take these data sources and use them in composite graphs to plot performance of an entire application (made up of multiple servers). At some point, I'll also add SNMP polling to pull in data from non-Orca data sources, but I haven't had much luck getting the polling setup.

Any insight or advice to point me in the right direction would be greatly appreciated.

Thanks,

Craig

[bbice]

Hmmm.... I think I'd let cacti create the RRD files where it wanted to (so I didn't have to go mucking about with paths to RRD files all the time), but copy the orca RRD files over to where cacti expects to find 'em. That's assuming, of course, that orca and cacti are monitoring the exact same data sources, named the same way, with the same min/max values, etc.

That would also ensure that you didn't have cacti and orca both modifying the same rrd files at the same time (which could be really painful to troubleshoot if graphs suddenly break).

I had a similar issue -- several machines setup with Munin plugins (these are production machines that it takes an act of <deity> to make any changes to - grin) but I want to graph the same data in cacti and move away from Munin. So I wrote a perl script that will connect to a machine running the munin plugin and query it, returning the data in a format cacti likes. I still have to add data input methods and templates for each munin plugin I want to query, but the script acts as the bit of glue between the two.

Brent