Notifications on the last 1 hour average

Hello,

you can easily do this by creating a Threshold-Trigger for this ping-sensor. This trigger should be set on the packet-loss channel and have a latency of 3600 (seconds), and the condition should of course be above 5%.

Best Regards.

Ahh, thanks for the clarification! I thought the latency means only that if the channel is over the specified threshold level during the specified delay time, then it triggers the alert. (like with the state triggers) If I understand you right then the delay is more a "summary period" in this case. Am I right?

Best regards!

The latency works similar with State and Threshold triggers. It always sets the time how long a condition (for example: DOWN; or ABOVE 5%) needs to be present before a notification will be triggered.

With a ping sensor I'm sending eg. 10 ICMP Echo requests. If all those 10 packets are failed, then the sensor is DOWN. If the sensor is DOWN for 1 hour, it triggers a notification. That's clear.

But what if once I loose 2 packets, then I loose no packets, and then again I loose 3 packets? If my packet losses follow the same pattern in the next 1 hour (loss, no loss, loss, etc.), will it trigger a notification if I have set up my threshold trigger to 5% with a delay of 3600 sec?

So does it summarizes the last one hour?

Thanks!

No, it would have to be a 5% packet loss for each scan. It does not work with averages.

Ok, so thanks for the reply. To be honest in real-life scenarios such threshold is not working well. I know from experience that if there's a packet loss, high latency, etc. to a remote device, then it will never be that clear to see.

If I set a notification to trigger with every occurrences, then it will definitely produce a tremendous amount of false alarm. If I set a threshold, then it's a possibility to miss a real issue just because there was 1 good result in the middle of a the successive bad results.

It would be nice to improve that part of PRTG.

Thanks, and happy new year!

You can acomplish this with a Custom Sensor that would parse the results of your ping sensor from the last hour.

• Gather your ping sensors data with the historicdata.csv api call.
• Loop through the result and determine the total packet loss
• Calculate the % packet loss and make this your sensors output.
• In the channels tab of your sensor set the limit to 5 (%) to alert you when the packet loss of the ping sensor goes over 5%

Hi Gerard,

Thanks for your reply. That's the kind of constructive thinking I'd like to see. I'm not sure if you are referring to the XMLValue custom sensor from PTF?

If so, then my concern is that in the API call for the historical data you have to specify the start and the end time, and I'm not sure how could I specify it dynamically?

Or do you mean writing a custom sensor by myself?

Thanks!

Hi Karoly,

The steps I mentioned were purely intended as "instructions" on how to achieve this with a custom sensor.

If you do not have the means of writing one yourself, I could probably have a look at it for you.

Gerard,

I have a very similar question or request. I am also looking to dynamically monitor average data for a few sensors over the past 1/2/4 hours etc. Do you have an example using the historicdata.csv api call that you could share and I could format for my needs? For example - have 1 sensor that monitors the past 1 hour averages for 4 other sensors and goes into a down state if any of them are over X% or amount.

Appreciate the help!

PS. The historical average or alert amount would be a great addition to a regular sensor!

Gerard you rock! Using a combination of this new custom sensor and using the sensor factory to get all values that I am looking to average this will be a cake walk to put into production with minimal setup on my end (always a plus). Thanks again!

Does someone know if the script PTF.SensorAverageXML still work with the current version of PRTG ? This is exactly what i need but I can't get it work : "Unable to connect to the remote server" Thanks!

Hi lporier,

Short answer, yes it does.

Please make sure that for the -ps= parameter you are using the exact same url when connecting to your PRTG server with your internet browser, including protocol and optional port number.

Like:

-ps=http(s)://myprtgserver(:8080)

A new version of this sensor, supporting TLS 1.1 and 1.2, is available at http://prtgtoolsfamily.com/downloads/sensorsxml

Thank you, it's working now, I'm able to get values.

@Torsten Lindner | PRTG Tools Family: this sensor is wonderful, thank you very very much!