10 Gigabit Ethernet shows as speed 4294967295

[evilzardoz]

I suppose my choices are one of two, as follows:

* Modify the template we're using for our 6500s to use ifHighSpeed as opposed to ifSpeed and (somehow?) multiply the value.

* Wait for this magical patch(?)


I've already been creating graphs with 10G limits as I wanted to allow for upgrades of links to 10G without manually tuning the graphs.

I can't imagine that we're going to exceed 4.2Gbit/sec in the near future, however I do anticipate that we will eventually.

[TheWitness]

Ok, so I checked the code for 0.8.6i, and the code that I had added to calculate the maximum value of the RRD was gone. This actually is a good thing in that all you have to do is add the ifHighSpeed to the XML file to use it on the graph. However, you do have to change the Data Templates "maximum" value to be 10Gig.

TheWitness

[jrichardson]

Was your template ever updated to use ifHighSpeed? If so, can you post your template for the rest of us? I have a couple 6500's with 10GB cards in them and have run into this issue too.

[styrofoam]

And is the fis really just running the rrdtool tune --maximum against the associated files, or is there intrinsically more to it?

Is there a way to search for .rrd files that have a ds[traffic_in].max = 4.2949672950e+09
in their properties, or is this a scripting exercise left to the reader?

[jrichardson]

And is the fis really just running the rrdtool tune --maximum against the associated files, or is there intrinsically more to it?

Is there a way to search for .rrd files that have a ds[traffic_in].max = 4.2949672950e+09
in their properties, or is this a scripting exercise left to the reader?

It wouldn't be hard to do from the cmd line.

for FILE in 'ls /path/to/rrds/' ;
do
rrdtool dump ${FILE} | grep 4.294

.... ect ....
done;

[styrofoam]

Ok, I'll just keep on asking the stupid questions. \


So I'm using rrdtoo tune $FILE --maximum traffic_in 1.0e+10

And now doing an 'rrdtool info' on that file spits out 'NaN' for the traffic_in variable maximum. Do I just need to wait for a poll cycle, or have I done something incorrectly?

Followup:
Yes, I did something incorrectly: traffic_in:1.0e+10 gives the results as expected. I'm apparently having issues decyphering the documentation.

[jrichardson]

So I'm using rrdtoo tune $FILE --maximum traffic_in 1.0e+10

I'm not 100% sure, but I dont think you can specify the number like that. I would try either:

Code: Select all

rrdtool tune ${FILE} --maximum traffic_in:1000000000

or you can tell it to ignore the max val with:

Code: Select all

rrdtool tune ${FILE} --maximum traffic_in:U

[styrofoam]

My next question: Does having the wrong maximum break all reporting of bits over maximum, or does it only break when you're running a graph with 95th percentile?

If cacti is simply reading the 64 bit octets in counter and throwing that on a graph, why does it care what the traffic_in maximum is? Does this have to do with the RRD database or something else entirely?

My problem here is obviously that I know enough to be dangerous, but not enough to be useful. (Apologies in advance.)

[jrichardson]

My next question: Does having the wrong maximum break all reporting of bits over maximum, or does it only break when you're running a graph with 95th percentile?

I don't know about 95th percentile. If the maximum is set too low, the RRD value will not store values > max in the database. Off the top of my head, i don't see how it would affect 95th percentile though. One risk of having it set too large is that if an anomoly occured where the value was 20Gbit for example, that would be stored in the database, even tho it was out of range.

If cacti is simply reading the 64 bit octets in counter and throwing that on a graph, why does it care what the traffic_in maximum is? Does this have to do with the RRD database or something else entirely?

It has to do with the rrd database, how cacti tells rrdtool to create it.

[styrofoam]

The only reason I said 95th percentile is that I imagined that the math would break with values greater than maximum. But a sanity check before a value gets entered into the database makes sense, thanks.

[sys0p]

I was reading closely this topic and it seems I am stuck in the same place as well, I have tried manual tuning :
rrdtool tune {RRDFILE} --maximum traffic_in:U

rrd_version = "0003"
step = 300
last_update = 1168807808
ds[traffic_in].type = "COUNTER"
ds[traffic_in].minimal_heartbeat = 600
ds[traffic_in].min = 0.0000000000e+00
ds[traffic_in].max = NaN


and I still can't get the maximum on this Te interface.

Can you please enlighten me what else I could try out ?

[gandalf]

I was reading closely this topic and it seems I am stuck in the same place as well, I have tried manual tuning :
rrdtool tune {RRDFILE} --maximum traffic_in:U

rrd_version = "0003"
step = 300
last_update = 1168807808
ds[traffic_in].type = "COUNTER"
ds[traffic_in].minimal_heartbeat = 600
ds[traffic_in].min = 0.0000000000e+00
ds[traffic_in].max = NaN

This is all correc to deactivate clipping of values during "rrdtool update". So rrdtool will simply store the data you give him.

and I still can't get the maximum on this Te interface.

Can you please enlighten me what else I could try out ?

Please describe in more detail. A screenshot, example with numbers etc. may help

Reinhard

[sys0p]

First of all I would like to thank you Reinhard for your fast feedback on this.

Basicly what I am trying to accomplish is to have Monthly 95th percentile 5 minute average graphs on TE interfaces .

I have tuned the rrd accordingly to what was posted here and I still cant seem to get over 4.4G .

Router interface stats sometimes show even 5Gbps and I cant see this in the data gathered by cacti..

rrd_version = "0003"
step = 300
last_update = 1169057405
ds[traffic_in].type = "COUNTER"
ds[traffic_in].minimal_heartbeat = 600
ds[traffic_in].min = 0.0000000000e+00
ds[traffic_in].max = 1.0000000000e+10
ds[traffic_in].last_ds = "1454639697250964"
ds[traffic_in].value = 2.5549986530e+09
ds[traffic_in].unknown_sec = 0
ds[traffic_out].type = "COUNTER"
ds[traffic_out].minimal_heartbeat = 600
ds[traffic_out].min = 0.0000000000e+00
ds[traffic_out].max = 1.0000000000e+10
ds[traffic_out].last_ds = "276851917596816"
ds[traffic_out].value = 5.5078984092e+08
ds[traffic_out].unknown_sec = 0
rra[0].cf = "AVERAGE"
rra[0].rows = 700
rra[0].pdp_per_row = 6
rra[0].xff = 5.0000000000e-01
rra[0].cdp_prep[0].value = 1.0029079190e+09
rra[0].cdp_prep[0].unknown_datapoints = 0
rra[0].cdp_prep[1].value = 2.1633830807e+08
rra[0].cdp_prep[1].unknown_datapoints = 0
rra[1].cf = "AVERAGE"
rra[1].rows = 11000
rra[1].pdp_per_row = 1
rra[1].xff = 5.0000000000e-01
rra[1].cdp_prep[0].value = NaN
rra[1].cdp_prep[0].unknown_datapoints = 0
rra[1].cdp_prep[1].value = NaN
rra[1].cdp_prep[1].unknown_datapoints = 0
rra[2].cf = "AVERAGE"
rra[2].rows = 775
rra[2].pdp_per_row = 24
rra[2].xff = 5.0000000000e-01
rra[2].cdp_prep[0].value = 1.0029079190e+09
rra[2].cdp_prep[0].unknown_datapoints = 0
rra[2].cdp_prep[1].value = 2.1633830807e+08
rra[2].cdp_prep[1].unknown_datapoints = 0
rra[3].cf = "AVERAGE"
rra[3].rows = 797


I have read another discussion based on the Monthly 95th percentile 5 minute average where it was explained that in order to do this you need to resize your current rrd to 8640 datapoints (288x30) or change the RRA definition accordingly.
Please let me know what I could be doing wrong, I am runing the latest cacti patched up to date.

Now I have two issues:
-Te interfaces to be able to graph more than 4.4Gbps
-Monthly 95th percentile 5 minutes average - to get more accurate figures instead of the normal way of cacti doing it now with 2 hour average.

I hope you will find the time to shed some light on this (:

Thank You !

[gandalf]

First one: perhaps an php number limitation? Not sure.
Second one: Yeah, you might resize. But AFAIK from my math lessons, that would not change the average
Reinhard

[sys0p]

Second one: Yeah, you might resize. But AFAIK from my math lessons, that would change change the average

I am not sure I got this correctly, what exactly you mean ?
I am a bit surprised why hasnt nobody documented this Monthly 95th percentile 5 minute average properly.. I think alot of people try to accomplish this without alot of luck..

Thank You !