Enter Case number reference for associated cases.

Case #

Status:

Type: APM as a Service

Old Article ID: 3571

Old Article Type: diagnostic

 

Include any information that is for INTERNAL use only

 For more information on "Peer Killer" and "Suspect" PP/LM/Mobile tests, see the following article:

INTERNAL Peer Killer and Suspect Tests

Information:

Detail the contextual information specific to the issue; i.e. Product, Version, Agent, System, etc.

  • Synthetic Classic Portal
  • Windows Recorder

Describe the problem, from the user perspective

I have received a message from Dynatrace saying that one of my Synthetic Classic tests was turned off:

It has come to our attention that the following test(s) have been automatically deactivated by our systems per our Quality of Service policy, because they were failing at or near 100%, and due to the nature of their script failures, if left active in their current state could possibly impact the availability of our (and your) testing peers:

[Account and test information]

We would be pleased to escalate this case to our Tier II Test Engineering team for further investigation and correction of the script(s); please confirm whether these test(s) are actively being utilized by your site at this time, or if they may remain inactive.

If you are not the primary administrator of these tests and would like us to contact someone else within your organization, please let us know.

We thank you in advance for your help.

What does this mean? Why was the tests inactivated?

Clearly list the Steps to resolve the issue

As part of Dynatrace's commitment to data quality, we proactively identify and inactivate Private Last Mile, Last Mile, and Mobile tests that are malfunctioning in such a way that they return invalid data; or they lock up the Peers, which prevents them from running other tests (thus decreasing the overall fill rate and availability of our Peer network).

For the tests identified as failing in such a way, looking at the listed availability for the tests in the Synthetic Classic Portal can be misleading, since Private Last Mile, Last Mile and Mobile calculates "failure percentage" differently from Backbone.

For Last Mile, Private Last Mile, and Mobile tests, "failure" is defined as the system's inability to successfully get results for a given test.

If a test is sent out to the Peers and valid results are not returned (i.e. either the data returned was earmarked as invalid, or the test shut down the Peer so that no other tests could be run against it), it is considered a failure. If, on the other hand, results do come back, it is considered a success even if the test itself failed.

If a test is only returning 5% of its measurements, it has a 95% failure rate according to our internal systems. However, if of that 5% of measurements returned, only 50% are failed tests, the Portal would show the Availability at 50%, making the test look as if it is failing less than it actually is.

It is important that these tests be corrected before being reactivated, either by you or by our Test Engineers, to ensure that you receive valid, useful data, and your Peers and fill rates are not negatively impacted by a malformed script.

Note the underlying reason for the problem

Malformed script affects data quality or Peer fill rates, and the test is proactively inactivated so that it can be corrected.