2

We are using Micro Focus UFT developer 15.01. About 1-2% of our tests never end, because they are stuck inside

Reporter.GenerateReport();

The issue is rare, but for us it is annoying enough to re-trigger our tests manually.

This is not our code. Anyone out there knows what could cause this issue? As far as I can tell, any test (failing or passing) can be affected, hence I assume it happens regardless of the content of report.

Corresponding post in Micro Focus community

More details as requested by Adelin:

  • Reporter.GenerateReport() is called once per TestSuite
  • Is the report folder generated even if the call doesn't finish? - Sorry, I don't know since test system just kills the VM after 1,5h.
  • The test is executed remotely on our internal cloud.
  • Debug level is Release. We never checked any other level.
tomwaitforitmy
  • 509
  • 3
  • 16
  • is `Reporter.GenerateReport()` called after every test or after every test suite? is the report folder generated even if the call doesn't finish? is the test executed remotely or locally? what is the debug level? does it happen at any debug level? please include more information. – Adelin Mar 17 '22 at 09:08
  • I added more information. We now wrapped the call with a TPL Task waiting for 2 minutes. If the call does not end, we skip attaching the report. What logs of UFT developer could we generate/attach in that case to help debugging? – tomwaitforitmy Mar 17 '22 at 14:06

1 Answers1

1

I will post the same as I did in the corresponding Micro Focus post:

The issue is hardly reproducible since it occurs in roughly 1-2% of our tests. We wrapped the line of code with:

 var success = Task.Run(() =>
 {
      reportPath = Reporter.GenerateReport();
 }).Wait(2*60*1000); //let's wait 2 minutes for the report maxCode

This prevents endless timeouts we were observing. Furthermore, in case of failure we can decide what do to. With only very few tests being impacted, we have not yet observed a single case where the test failed and the report was not generated.

We assume that some process (such as virus scanner) on our VMs is causing this issue. If there are any log files we could collect in such cases, that would be a possibility to investigate further. We are, however, satisfied with our solution.

tomwaitforitmy
  • 509
  • 3
  • 16
  • Well, you hace no solution. You just have a workaround to eliminate the wait, and get no report in return. – TheBlastOne Feb 14 '23 at 09:07
  • 1
    I would be happy to accept any better solution. You´re right: It is just a workaround, but still better than having ~2% of your tests failing each night for no apparent reason. – tomwaitforitmy Feb 14 '23 at 09:22
  • 1
    I consider this a serious bug. We tracked it down to being some sort of sporadic symptom caused by some internal issue. There is nothing you can do about it. Some tests hang during "Uploading results" for HOURS. If you wait long enough, the run result will indeed get uploaded. We´ve seen uploads completed 8.5 hours after script execution (even if empty, or "hello world") has ended. It has nothing to do with the component (i.e. script) contents, even a blank test script can trigger this symptom. Very, very unfortunate that this has not been taken seriously and not fixed in the last 6 years. – TheBlastOne Feb 14 '23 at 09:50
  • Indeed, I have similar observations. – tomwaitforitmy Feb 14 '23 at 14:03