@rickyma

Test Results

 2 657 files  ±0   2 657 suites  ±0   5h 32m 49s ⏱️ +22s
   946 tests ±0     945 ✅ ±0   1 💤 ±0  0 ❌ ±0 
11 799 runs  ±0  11 784 ✅ ±0  15 💤 ±0  0 ❌ ±0 

Results for commit 93151aa. ± Comparison against base commit 64890eb.

This is not from log4j2. I don't think log4j2 will hang.

This is not from log4j2. I don't think log4j2 will hang.

Is this right? I'm not sure. Please see https://logging.apache.org/log4j/2.x/performance.html#asynchronous-logging-with-caller-location-information

Your jstack log does tell it's from log4j2. Instead, it should belong to log4j1.

We don't need the slf4j-reload4j dependency, we can remove it. That's probably the reason why it hangs.

Yes. I know this cause that not happen on the log4j2. But the doc indicates that we should not use the Method in the log style

We can remove it. When debuging, we can add it back manually.

Or we could just add something like a debug switch, then we can use different log4j2 configurations in different scenarios (test/prod).

We can remove it. When debuging, we can add it back manually.

Or we could just add something like a debug switch, then we can use different log4j2 configurations in different scenarios (test/prod).

Ok for me. I just hope the example log4j xml could be stable and available

You can modify the description of the issue #1910, because the jstack log has nothing to do with this PR. This PR will not solve the issue.