Verify on Collectd 5.0.x

[indygreg]

Collectd 5.0.0 was released recently. Need to verify that plugin works there.

[rogerger]

Hi, We are testing on Collectd 5.0.0 but we are having issue with derive value. The network if_octets is about 10 times as much as what it is supposed to be: The following is the comparison between rrd and whisper:

Actual Value from cgp(rrdtool):

1304473100: 1.0157142857e+01
1304473170: 8.7571428571e+00
1304473240: 1.0057142857e+01
1304473310: 7.6714285714e+00
1304473380: 9.6000000000e+00
1304473450: 1.1685714286e+01
1304473520: 1.1371428571e+01
1304473590: 1.2242857143e+01
1304473660: 9.7571428571e+00
1304473730: 8.6571428571e+00
1304473800: 1.0442857143e+01
1304473870: 1.1714285714e+01
1304473940: 9.3571428571e+00

Actual Value from graphite(whisper):

1304473140 109.000000
1304473200 114.000000
1304473260 45.000000
1304473320 86.000000
1304473380 140.000000
1304473440 113.000000
1304473500 105.000000
1304473560 72.000000
1304473620 81.000000
1304473680 86.000000
1304473740 68.000000

Do you see this problem with 4.x? Thanks.

[indygreg]

Well, if you are seeing values on Collectd 5.0, I guess that means the plugin loads, which is good news! Now, to address your issue.

What's interesting is that the reported values are off by about 10x. My theory as to what's happening is your Collectd's polling interval is 10 seconds. And, RRD is storing the actual derivative value (dx / dt) as opposed to the change in value (dx), which this plugin records. I think that if you ask RRDTool to graph your data, it realizes it is stored as derivatives and performs an integral-like operation to transform it into actual values.

The question that really needs answered is whether the values reported by this plugin are correct or whether they are off by a factor of polling interval times.

[Dairenn]

Polling interval times are a very likely culprit. According to http://graphite.wikidot.com/faq#toc8:

Data gets put into a temporary area inside the RRD database where it is not accessible until the current time interval has passed and another value is inserted into the database for the following interval. If that does not happen within an allotted period of time, the original data point will get overwritten and is lost. Now for some metrics, the lack of a value can be correctly interpreted as a value of zero. However, this is not the case for metrics like latency because a zero indicates that work was done in zero time, which is different than saying no work was done. Assuming a zero value for latency also screws up analysis like calculating the average latency, etc.

[TimidRobot]

Works with Collectd 5.0.1

I am also seeing values off by 10, but it looks like this is already covered by #8 and #9.

[indygreg]

Thanks for the verification! I'm going to close this bug per your comment.