Y-axis phenomenon

tkis · Post by **tkis** » Thu May 10, 2007 9:21 am

Hello everybody....in the last 2 days I came across to a very weird thing which I cannot explain. Let me give you a brief overview:

I have two identical Cisco routers (Router1 & Router2) and I monitor the same OID (gauge) in two different graphs. The first graph shows the correct scale at y-axis and the second graph shows some strange values like 10m.The expected values are pure integers like 1, 2, 3, etc.

I have enabled in both the "Auto scale" and still nothing. The two configurations are identical. The only difference is the IP address.
I delete the problematic configuration and I did it again from the scratch but I have the same respond.
The template tha I am using is "SNMP-Generic OID template".

Any ideas???????????
I am trying to solve this problem for the last two days and unfortunately I cannot explain this phenomenon.

Thanks in advance!

http://www.blwebhosting.com.cy/

BSOD2600 · Post by **BSOD2600** » Thu May 10, 2007 11:12 am

Read http://forums.cacti.net/viewtopic.php?t=8168 for what 10m means.

tkis · Post by **tkis** » Fri May 11, 2007 2:23 am

ignore

tkis · Post by **tkis** » Fri May 11, 2007 2:24 am

ignore

tkis · Post by **tkis** » Fri May 11, 2007 2:25 am

ignore

tkis · Post by **tkis** » Fri May 11, 2007 2:25 am

Thanks for your quick reply.
I know what 'm' means. The problem that I am facing is that I had to identical nodes with two identical graphs and one of them shows wrong y-axis values.
I run the 'snmpwalk' command in Linux:

Snmpwalk --On --c public 1 <ip@> <OID>

<OID> = Gauge: 2

This means this <OID> respond back with an integer.

See attach graphs for more clear view. If you see at the graphs the green one is the correct, and thats how the red one should look like. At the green one I have two connections and stay at a constant line in all the time range that I was connected. However, in red one graph I have two pulses at the time I was connected, with y-axis showing a wrong value.

Any ideas why this is happening?

Post by **gandalf** » Fri May 11, 2007 9:56 am

I'm quite sure the second one is a graph of a COUNTER type data source whereas the first one shows a GAUGE one. Please perform "rrdtool info" on both rrd files to verify
Reinhard

tkis · Post by **tkis** » Wed May 16, 2007 1:35 am

Hello,

Unfortunately both of them are GAUGE. Find attach the "Data Source Debug" output for the above graphs.

If I understand correct "rrdtool info" produce a file which can you find in the attach file.

Is that correct?

Thanks in advance,
Tkis

http://www.blwebhosting.com.cy/

Post by **gandalf** » Wed May 16, 2007 2:49 pm

This was not exactly what I asked for but it holds the same information. I'm quite sure that you've changed rra definitions and added a new item. But they are in conflict and may produce unpredictable behaviour. Please post a screenshot if you are in doubt which one to remove.
And you will have to recreate those rrd files or at least will ahve to export, change and re-import
Reinhard

tkis · Post by **tkis** » Thu May 17, 2007 2:03 am

Hi,

Can you please explain in more detail. I don't know which of the two I have to delete. I have post screen shots. I am not pretty sure if the below screen shots are the requested but if you need more screen shots let me know.

Many thanks!

Post by **gandalf** » Thu May 17, 2007 11:55 am

Wrong, please use http://localhost/cacti/rra.php
Reinhard

tkis · Post by **tkis** » Fri May 18, 2007 4:19 am

Hello!

Is the below screen shot that you ask for?

If not, can you please give me more details of what are you looking for?

http://www.blwebhosting.com.cy/

Post by **gandalf** » Fri May 18, 2007 8:17 am

Yeah, that's the right one. You see, there are two rra's which hold a "Step" of 1. This is not permitted. The step level always defines an aggregation/consolidation level and determines, how many "original" data points should be consolidated for that very level. The first level usually holds step=1 (that is: this level holds un-consolidated data). If you're running the 1 min poller patch, the first line may be fine but the second is weird. If you're running the default 5 min polling interval, the first line is obsolete. Please decide
Reinhard

tkis · Post by **tkis** » Mon May 21, 2007 2:23 am

Thanks for your help!
Now that you found the problem what is the solution? I am currently monitor more that 300 graphs from Cacti in our network. Consequently, some of them are using the 1 min poller patch (Hourly -1 Minute Average) and the rest of them are using Daily (5 Minute Average). Base on that I have some question for you:

1. If I delete one of them what are the consequences will have in the existing graphs?
2. Do we have any alternative solution? And if yes can you post it step-by-step? I prefer to not delete any of them.
3. For the above posted graphs I prefer to use the 1 min poller patch (Hourly -1 Minute Average)?

Thanks in advance.

http://www.blwebhosting.com.cy/

Post by **gandalf** » Mon May 21, 2007 2:24 pm

To be honest, I can't answer the question as I have no knowledge of the 1 minute polling patch. Please post your question to that thread to take advantage of experiences more knowledgeable users have gained using this patch
Reinhard

Y-axis phenomenon

Y-axis phenomenon

Who is online