shaonan666 commented on issue #3569:
URL:
https://github.com/apache/logging-log4j2/issues/3569#issuecomment-2754340901
I think I’ve got it. Let me describe what happened again to ensure I fully
understand.
**My program doesn’t specify a `messageFactory`.**
In version 2.23.1, `LoggerContext#getLogger(String name) `ultimately
initializes a `ParameterizedMessageFactory`
https://github.com/apache/logging-log4j2/blob/fea2a7116160fb1555d578406444b4fc4f0ef2da/log4j-core/src/main/java/org/apache/logging/log4j/core/LoggerContext.java#L493-L495
https://github.com/apache/logging-log4j2/blob/fea2a7116160fb1555d578406444b4fc4f0ef2da/log4j-api/src/main/java/org/apache/logging/log4j/spi/AbstractLogger.java#L134-L137
In version 2.24.3, `LoggerContext#getLogger(String name) ` uses a
GC-free-compliant `ReusableMessageFactory` based on the configuration
`log4j2.enableThreadlocals=true`
https://github.com/apache/logging-log4j2/blob/4aa5efec7f2ee2d4623381cd66c3df3a16a6b9a9/log4j-core/src/main/java/org/apache/logging/log4j/core/LoggerContext.java#L93-L97
https://github.com/apache/logging-log4j2/blob/4aa5efec7f2ee2d4623381cd66c3df3a16a6b9a9/log4j-core/src/main/java/org/apache/logging/log4j/core/LoggerContext.java#L538-L540
These two different MessageFactory implementations are the main reason I’ve
observed a gradual increase in memory usage: to support GC-free,
`ReusableMessageFactory` allocates objects that aren’t released until all
`RingBufferLogEvent` instances have been used at least once, at which point the
memory growth stops.
---
>Do you expect Log4j Core to pre-allocate StringBuilders of size
[log4j2.initialReusableMsgSize](https://logging.apache.org/log4j/2.x/manual/garbagefree.html#log4j2.initialReusableMsgSize)
instead of doing it on-demand?
Pre-allocating a suitably sized `StringBuilder` based on prediction could be
a good choice for services with high runtime latency requirements. But my main
concern is the gradual memory increase in version 2.24.3, which I didn’t see in
2.23.1, and I initially suspected a memory leak. However, based on the current
situation, it doesn’t seem to be a memory leak risk. I’m considering reducing
the RingBuffer size to manage memory usage.
Thank you for your patient explanations!
--
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: notifications-unsubscr...@logging.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
