Increase timeouts to fix flaky ServiceFabricProcessor test cases #7446
Conversation
JamesBirdsall
requested review from
jsquire,
kinelski and
serkantkaraca
as code owners
|
There are some tests still failing under the Checks tab. |
Yes, there's one case which has a separate issue. I have re-disabled it and am waiting for the checks to run again. |
| @@ -102,7 +102,7 @@ public void VerifyNormalStartup(int maxRetries) | |||
| int retries = 0; | |||
| while (!this.Processor.IsOpened && !this.HasShutDown && (retries < maxRetries)) | |||
| { | |||
| Thread.Sleep(1000); | |||
| Thread.Sleep(5000); | |||
There was a problem hiding this comment.
We should really try to find a better solution then adding additional sleep time, all that does is end up making our test runs take longer and still have reliability issues. /cc @mikeharder
There was a problem hiding this comment.
It appears the intention of these APIs is to block until a desired state is detected and fail after some timeout. If so, these APIs should take TimeSpan maxDuration rather than int maxRetries, and they should poll as frequently as reasonable (say 100ms instead of 1000ms or 5000ms). This gives the API sufficient time to pass reliably but ensures the API returns as quickly as possible.
Multiple cases, including the ones marked as flaky, are failing because the processor does not start up as quickly as expected. All cases work fine on my desktop, but it is not surprising that a loaded common testbed would be slower. Increased the timeout.