Cactiusers.org

I have huge problems with thold and "nan" values in RRDtool databases.

1) If any DS of a Cacti Data Source is "nan", the thold plugin can't fetch any value of the RRDtool database, even if some RRDtool DS are not "nan". In this situation, when I create a threshold, thold does not display the value of any DS.

Example: Data Source with two DS
DS1 is currently 100
DS2 is currently nan.

Thold does not display the data for DS1 nor DS2. So, if I have set a (low) threshold on DS1, thold will trigger a wrong alarm if DS2 is nan and DS1 is okay.

I have looked in the cacti log. The rrdtool fetch command is okay. And if all DS are not nan, thold works as expected.

2) If a low threshold is set and the current DS is nan, thold does initiate a threshold alarm. This is a minor "bug", which was already reported in "Feature requests". I think, this behavior is related to bug No. 1, since it seems that thold can't handle "nan" at all.

System Info:
Cacti 0.8.6h
Thold 0.2.7
(problems occur also with 0.8.6g and 0.2.6)

I would appreciate any help on solving this issue.

I have actually been debugging this problem for awhile already, the issue is that the rrdtool_execute doesn't seem to be properly returning the value if there is a nan. I kinda lost track of where I was on this bug for a while as the datasource that I was having the issue with fixed itself after a re-install on my test machine. I will see if I can cause another datasource to break so that I can start testing fixes for this again.

Thanks!

Great. If you need any help for testing, please let me know.

This bug is very important for me, because almost all my RRDtool databases are having DS with nan values occuring quite often. In addition, I have also some spare DS that I don't update with values all the time. So, thold does currently not work for these databases.
I have also read some posts on cacti.net about people having observed that thold sometimes does not view the current value on the thold page. I guess that these problems are caused by this "nan bug" also.

Thanks again

Yep, they are one and the same. I see what I can do about it.

Any update on this? I'm seeing the same issue.
Thold 0.3.0
cacti Version 0.8.6i

I'm running Windows.

Let me know if I can help troubleshoot.

Its not an issue anymore in the current version.

The latest version is 0.3.0 correct? I'm still seeing this problem. Maybe I'm doing something wrong? What info can I give to troubleshoot?

Sorry, I meant the latest SVN doesn't have this issue anymore. I promise to get it out eventually, just still adding a few items. For instance, a few minutes ago I added the ability to Create Thresholds for multiple hosts at once by using the drop down box on the Devices page.

Awesome. Sounds like some great features are coming. I look forward to it.

I really appreciate the work that you put into this.

Let me know if there is anything I can do to help

I have the same issue.

Any solutions?

Regards

I have compiled a short document about all changes I have made towards thold 0.3.0. See PDF doc attached.

For this specific bug I have changed the function rrdtool_function_fetch() in rrd.php.

Fix regex matching of nan values on all DS positions
Old (line 401-403):

New: Just accept all non-whitespace characters between spaces (incl. nan)


In addition you have to change the function call in thold-functions.php:
in order to show unknown values.

See attached doc for more info about other NaN bugs I have fixed. For me thold 0.3.0 works now reliable, especially since I can't use newer releases, because I don't update my RRD files with the poller.

Hi ralfk

The file that you posted worked fine, just for hdd_total, but hdd_used and hdd_percent still not working, the value is "U".

Do you have any idea what's happening?

Regards

My next release will have an option to switch back to the old way (hopefully). From a technical standpoint, it should be an easy swap, but we will see how it goes.

That would be really great !

Sorry, no idea. You would need to provide much more details about your problem (sample rrd file with values, what values and thold behavior do you expect ?)