On 08/10/2019 17:49, Mark Thomas wrote: > On 07/10/2019 17:40, Igal Sapir wrote: >> Mark, >> >> On 10/7/2019 4:51 AM, Mark Thomas wrote: >>> The proposed Apache Tomcat 9.0.27 release is now available for voting. >> >> I'm getting the failures below [1] for unit tests on Windows 10 with >> Java 1.8u181. False positives? > > Don't know yet. I didn't get the same failures on Windows but I was > using a different OS and a different JRE. > >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt: FAILED >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt-expected:<...3-Header-[:status]-[[304] >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt-3-Header-[etag]-[W/"957-1447269522000"]] >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt-3-Header-[date]-[W...> >> but was:<...3-Header-[:status]-[[200] >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt-3-Header-[accept-ranges]-[bytes] >> TEST-org.apache.coyote.http2.TestStreamProcessor.APR.txt-3-Header-[etag]-[W/"957-1447269522000"] > > This is TestStreamProcessor.testPrepareHeaders() failing for each connector.
I think the machine you are using for testing has some sort of issue with its clock. The above test will fail like that if the last modified time of the file being requested is ahead of the current time. I don't see how that can happen - even if the tests have just been checked out - unless the clock moves back in time for some reason. Overall, I'm not concerned about these failures although I am interested to hear of you find out what is going on. Mark > > <snip/> > >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt: FAILED >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt-Socket.timeoutSet >> failed (<1s) [999760800] +-[400] >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt-junit.framework.AssertionFailedError: >> Socket.timeoutSet failed (<1s) [999760800] +-[400] >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt- at >> org.apache.tomcat.jni.TestSocketServer.testBlockingReadFromClientWithTimeout(TestSocketServer.java:111) >> >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt- >> TEST-org.apache.tomcat.jni.TestSocketServer.NIO2.txt-Testcase: testPort >> took 0.001 sec > > That is a false positive. The timing variance is significantly greater > than the error margin. It looks like we need to adjust that. Interesting > that the socket waited a little less than a second to timeout. We'd need > to increase the error margin by 3 orders of magnitude (400ns to 0.4ms) > for that test to pass. > > I'll start looking into these but more eyes are always welcome. > > Mark > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org > For additional commands, e-mail: dev-h...@tomcat.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
