Hi Leon,
Yes by default the thread name is cached.
There is a system property to switch that off: see
https://github.com/apache/logging-log4j2/blob/master/log4j-core/src/main/java/org/apache/logging/log4j/core/async/ThreadNameCachingStrategy.java
The reason for the caching is that every call to get the current thread's name
creates a new String object, which I found impacted performance measurably.
Why Thread is implemented to store its name in a char[] array and create a new
String on each query I don't know. May have to do with interaction with native
code.
Remko
Sent from my iPhone
> On Feb 16, 2017, at 4:37, Leon Finker <leon...@gmail.com> wrote:
>
> Hi,
>
> Is there any thread name caching going on with log4j2 (doesn't matter
> sync/async)? If I have a thread pool of tasks and call
> Thread.currentThread().setName (with unique id) in the beginning of the task
> execution, I sometimes see reuse of previous thread name in the log
> statements. Id is UUID, so it's definitely unique. And if I do System.out on
> the current thread name after setting it, I do see the new name as expected.
> But in log statements the thread name is from the previous set. The pattern
> layout is: [%date{DEFAULT}{UTC}Z][%level][%thread:%tid][%c{3}] %message%n
>
> log4j:2.7
>
> Any ideas what could be causing the thread name reuse?
>
> Thank you
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org
> For additional commands, e-mail: log4j-dev-h...@logging.apache.org
>
