Thold gathers interface statistics incorrectly

[shep]

I am new to cacti and thold so I probably don't have something configured correctly. I wan Thold to send me an alert when an interface reaches 80% of its avaliable bandwidth in or out. The issue is that when I look at the treasholds the current traffic is less than 1 as in .00xx I have seen it go above this into more than 1 and I get an alert if the number higher than 80, but it does not seem to be at 80%.
screen shots




Any help would be greatly appreciated. Thanks

[cigamit]

The problem I see is this. You have the Percentage Datasource set to the same datasource as the Data Field, when instead it should be whatever datasource holds your Max traffic. You should probably state your Thold version, as my version doesn't allow you to select the current one.

Percentage = (current / max) * 100

Since you are doing (current / current) * 100, its interesting that it would ever by anything other than 100 (I shall have to look into that bug).

The default network traffic template doesn't graph the line speed, you will have to create a new template to do that. This is not the most ideal situation, considering the speed of the network card is usually what is pulled (10M, 100M, 1000M) and may not be the actual line speed (1.5M, 3M, etc....) so your percentage may not represent a true percentage.

[shep]

I only have traffic in or traffic out avaliabe to select as a datasource. How do I make another data source that is traffic max? As I said I am new at this. I am using Thold 0.4.1 on Cacti 8.7g

[cigamit]

Actually, it would be easier with a CDEF instead of using the percentages.

Goto the console and click on Graph Management. Under that you will now see a CDEF link on the left, click it. Create a new CDEF.

Name it "Traffic Percentage (100Mb)". You will then use the drop downs to create a CDEF that looks like this.

Code: Select all

Item            Item Value	 
Item #1         Special Data Source: CURRENT_DATA_SOURCE	  	
Item #2         Custom String: 13107200
Item #3         Operator: /	  	
Item #4         Custom String: 100	  	
Item #5         Operator: *

FYI: 13107200 = (100 * 1024 * 1024 ) / 8

Now that you have the CDEF. Goto back to your threshold template (or threshold) and change the Date Type to CDEF and select the one you created from the dropdown. You should now see it reporting as a correct percentage of a 100MB link. If you have different speed links, then just create separate CDEFS (duplicate is made for this) for each one and manually apply it to each threshold. All you need to change is the name and the 13107200 number above
(LinkInMb * 1024 * 1024) / 8 ).

I have a 10Mb version of this applied to a 10Mb link I have, and it seems to be working just fine so far.

[in2deep]

Please help! can some confirm if the network traffic_out result below is in bytes.

traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.2, value: 3844305579 (what is this??)

It doesn't make any sense to me since I have a gigabits NIC.
1Gitbits = 1073741824 < 3844305579 * 8 bits (convert to bits)

Thanks in advance.

Tho

[noname]

traffic_out, oid: .1.3.6.1.2.1.2.2.1.16.2, value: 3844305579 (what is this??)

It doesn't make any sense to me since I have a gigabits NIC.
1Gitbits = 1073741824 < 3844305579 * 8 bits (convert to bits)

Did you reboot that device or restart SNMP service of it?
I often happen to see these 'spike' traffics when doing so.

[in2deep]

Is not just this one host. I have multiple hosts with the traffic in/out with result value greater than 1073741824. Thanks.