afs commented on issue #2328: URL: https://github.com/apache/jena/issues/2328#issuecomment-1997327939
After running GH Windows actions repeatedly across the day and also running in batches of 3-5 at the same time, the test suite is less vulnerable with PR#2328 applied but not perfect. Usually, it is an auth test that fails; not always the same auth test. In an auth related test, there are two HTTP requests - the first attempt results in a 401 and the second with the authentication credentials. But it is not exclusively auth tests failing. With groups of tests running at the same time, other tests that are a single HTTP action fail sometimes. I'm guessing that a group of tests do share some underlying hardware so a batch changes the load on the server - this is consistent with the build action being more likely to fail at busy times of the EU-US working day. The common factor is creating a Fuseki Server in a Junit `@BeforeClass` and stopping it in `@AfterClass`. From previous Jena test stability experience, Windows behaves different to Linux in the timing of old ports being collected by the OS kernel and in the way that thread scheduling can cause longer pauses of a thread. That on its own isn't a deep explanation. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
