Unsolved
This post is more than 5 years old
3 Posts
0
735
Clarify use of MinimumErrorRate and ErrorTrafficThreshold to prevent HighErrorRate Events on interfaces with low utilization
InCharge SMARTS 9.2 native monitoring of interface errors has 2 constraints:
ErrorThreshold: The upper threshold for packet errors expressed as a percentage of the total number of packets.
ErrorTrafficThreshold: The minimum error traffic expressed as a percentage of the total bandwidth to generate the HighErrorRate event. ErrorTrafficThreshold accepts any value between 0.0 to 100.0.
According to this article https://support.emc.com/kb/10677
ErrorThreshold: This threshold is used as the UPPER threshold for triggering a HighErrorRate event.
ErrorTrafficThreshold value is used to calculate a MinimumErrorRate value which is in turn used as a LOWER threshold limit for triggering a HighErrorRate event.
It also notes that for Smarts IP to trigger a HighErrorRate event, the error rate received has to be above both the limits of the ErrorThreshold and MinimumErrorRate (calculated using the ErrorTrafficThreshold).
Would like to eliminate situations where interfaces have no real traffic and a single error causes an alarm to be thrown. I've read related articles about how HighErrorRate events are calculated but feel there's a gap in my knowledge on how to leverage these parameters so that HighErrorRate alarms are only generated when an interface is 5% or more utilized with packets (not error packets, all packets) AND there is 1:1000 errors ratio natively alarm. Am looking for clarification on how to leverage these (and possibly other parameters) to set the conditions I've described for generating HighErrorRate events.
PaulORourke
170 Posts
0
February 12th, 2015 02:00
Hi smartandincharge,
You are correct regarding the descriptions regarding the threshold setting above.
Setting the following ErrorThreshold should satisfy the 1:1000 errors ratio:
ErrorThreshold 0.001
However the ErrorTrafficThreshold is a measure of the error traffic vs the total bandwidth of the interface.
There is not have a threshold setting for the minimum normal traffic before an alarm is generated.
The main reason for using this method is to prevent false alarms when there is only minimal traffic on the interface, so a small number of error packets could quickly see the ErrorThreshold percentage exceeded. Smarts IP does not want to generate an alarm when the sampling size is small.
Using the ErrorTrafficThreshold as a lower limit and the ErrorThreshold as an upper limit, Smarts ensures that both of the following conditions are met before it will generate a HighErrorRate event:
The required minimum threshold for the number of error packets received (ErrorTrafficThreshold) has been exceeded and
the upper limit threshold for the percentage of errors (ErrorThreshold) has been exceeded.
Please let me know if you have any additional questions.
Kind Regards,
Paul O'Rourke
SmartAndInCharg
3 Posts
0
February 12th, 2015 08:00
Appreciate the reply Paul. Not sure that ErrorThreshold accepts fractional values. Heard that a version newer than 9.2.2 might support this but haven't found any technical literature that describes this. And like you mentioned, I'd still need to be able to define a minimum total traffic threshold to compare against before alarming.
Do you/anyone have a suggestion on how I can accomplish monitoring and alarming for the condition I described?
Thanks again.