I’ve seen errors like that before. I’ve also tried figuring out how to make 
these tests more robust, but they require overhauling the rolling appender code 
for testability and such, something that makes more sense as a larger effort to 
overhaul rolling appenders.

> On Sep 12, 2024, at 07:17, Gary D. Gregory <ggreg...@apache.org> wrote:
> 
> Any thoughts on these:
> 
> [INFO] Results:
> [INFO]
> [ERROR] Failures:
> [ERROR]   RollingAppenderDeleteAccumulatedSizeTest.testAppender:68 
> [target\rolling-with-delete-accum-size\test\test-5.log, 
> target\rolling-with-delete-accum-size\test\test-6.log, 
> target\rolling-with-delete-accum-size\test\test-7.log, 
> target\rolling-with-delete-accum-size\test\test-8.log, 
> target\rolling-with-delete-accum-size\test\test-9.log] ==> expected: <4> but 
> was: <5>
> [ERROR]   RollingAppenderDeleteNestedTest.testAppender:85 
> [target\rolling-with-delete-nested\test\my-1.log, 
> target\rolling-with-delete-nested\test\my-2.log, 
> target\rolling-with-delete-nested\test\my-3.log, 
> target\rolling-with-delete-nested\test\my-4.log, 
> target\rolling-with-delete-nested\test\my-5.log, 
> target\rolling-with-delete-nested\test\test-6.log, 
> target\rolling-with-delete-nested\test\test-7.log, 
> target\rolling-with-delete-nested\test\test-8.log, 
> target\rolling-with-delete-nested\test\test-9.log] ==> expected: <8> but was: 
> <9>
> [INFO]
> [ERROR] Tests run: 1555, Failures: 2, Errors: 0, Skipped: 7
> [INFO]
> 
> Gary
> 
> PS: git main at git log -1
> commit fda45ecf6858734136ff584544b76d558843a3a4 (HEAD -> main, origin/main)
> Author: Piotr P. Karwasz <piotr.git...@karwasz.org>
> Date:   Thu Sep 12 08:43:45 2024 +0200
> 
>    Increase timeout in `AsyncLoggerConfigTest`
> 
> G
> 
> On 2024/09/12 12:08:07 "Gary D. Gregory" wrote:
>> On 2024/09/12 06:51:22 "Piotr P. Karwasz" wrote:
>>> Hi Gary,
>>> 
>>> On Wed, 11 Sept 2024 at 22:13, Gary D. Gregory <ggreg...@apache.org> wrote:
>>>> Running 'mvn clean install' on git main [1] gives me:
>>>> 
>>>> [ERROR] Failures:
>>>> [ERROR]   AsyncLoggerConfigTest.testSingleFilterInvocation:114
>>>> Wanted but not invoked:
>>>> appender.append(<any>);
>>>> -> at 
>>>> org.apache.logging.log4j.async.logger.AsyncLoggerConfigTest.testSingleFilterInvocation(AsyncLoggerConfigTest.java:114)
>>>> 
>>>> However, there was exactly 1 interaction with this mock:
>>>> appender.getName();
>>>> -> at 
>>>> org.apache.logging.log4j.core.config.AppenderControl.<init>(AppenderControl.java:50)
>>> 
>>> In this test a message is sent through `AsyncLoggerConfig` and we wait
>>> up to 100 ms until `Appender.append()` is called on the asynchronous
>>> thread.
>>> I guess this _might_ fail on an extremely busy host. If your host was
>>> not so busy, we might have a performance problem with LMAX Disruptor.
>>> 
>>> I bumped the maximum delay to up to 1 s, hope it helps.
>> 
>> Thank you Piotr.
>> 
>> When I run a build, I run it and then do other work on my PC, so I expect 
>> the CPU usage to go and down a lot. It would be nice if we could find a way 
>> to use structures like count down latches to sync up some tests. The build 
>> is running now... ;-)
>> 
>> Gary
>> 
>>> 
>>> Piotr
>>> 
>>> PS: On a side note: `log4j-mongodb` might fail on some hosts (e.g. my
>>> Debian 12), since it requires some specific versions of OpenSSL and
>>> other libraries.
>>> That is why I proposed to start the Mongo server from Docker in [1].
>>> 
>>> [1] https://github.com/apache/logging-log4j2/pull/2922
>>> 
>> 

Reply via email to