External RRDs generated by NFSEN not displaying properly

[sayfuzzypickles]

Greetings!

I've been lurking the forums for awhile now and have yet to find someone who has encountered similar problems, and am looking for some ideas/direction.

Background:
I have a cacti instance running on a webserver (let's call it www) and I have nfcap/nfdump/nfsen running on another machine (we'll call it sense). Sense also has its own instance of cacti running. Both www and sense have the same version of cacti (0.8.8b) and rrdtool (1.4.8 ) running. We've set up an sshfs mount for www to be able to mount where sense is storing the rrds being generated by nfsen.

The problem:
On sense's cacti instance, reading in the external rrds (following the advice in http://docs.cacti.net/manual:087:8_rrdt ... ernal_rrds) works like a charm. The graphs are beautiful. When we use www's cacti to read in the data sources, problems arise. The graphs are being created, but the data is completely wrong. There are spikes all over the place, missing data, etc. If I refresh the page where the graphs are drawn, I can see the graph actually change randomly throughout the entire graph. Spikes will raise/lower, some data will disappear - it's odd to say the least.

My first thought was architecture differences. www is running 64-bit NetBSD, and sense is 64-bit Debian. As a test, I did an rrdtool dump <rrdFile> <xmlFile> on sense, scp'd it over to www, and did an rrdtool restore <xmlFile> <rrdFile> and had www use the new rrd file to create the graph (since they state in their documentation that this is architecture-independent): Still no dice.

Are there any glaring issues that I'm overlooking? I'm starting to run out of places to look (nothing apparent in the log files, etc). The fact that the dump/restore didn't work makes me wonder if this is an issue within Cacti, and not the actual rrd files. Any advice is greatly appreciated. I can provide additional info if needed.

Thanks!

[sayfuzzypickles]

*tap tap* Is this thing on?


Anyone?

[sayfuzzypickles]

Well, for what it's worth, I figured out what is happening. But I'm not sure why or how to prevent it from happening again.

The graphs are silently changing which data sources they are using. They all will change (seemingly at random) what data source is to be graphed, and will always pick the first one that is listed in the directory, essentially graphing the same RRD many times, causing the weird graphs.


Anyone ever run into this behavior? Going to be flexing my google muscles to try and sort this out.

[sayfuzzypickles]

Found the fix:

Each time you update the graph template, the data sources associated to the item you edited will change which data source it uses. So, once you have the graph template the way you want it, THEN go in and assign the data sources to be graphed. If you try to edit the template to see how it looks and expect to see legitimate results, you're gonna have a bad time.


I hope this helps someone in the future.

*fist bump*