Really god! We’re in the right way.
I’ve set the last cb dependencies based on master branch and the last memory leak is this threadDeathWatcher:
13:42:04.218 [cb-io-1-4] DEBUG c.c.c.d.i.n.buffer.PoolThreadCache - Freed 21 thread-local buffer(s) from thread: cb-io-1-4
13:42:04.218 [cb-io-1-3] DEBUG c.c.c.d.i.n.buffer.PoolThreadCache - Freed 10 thread-local buffer(s) from thread: cb-io-1-3
13:42:04.218 [cb-io-1-2] DEBUG c.c.c.d.i.n.buffer.PoolThreadCache - Freed 16 thread-local buffer(s) from thread: cb-io-1-2
13:42:04.219 [cb-io-1-1] DEBUG c.c.c.d.i.n.buffer.PoolThreadCache - Freed 13 thread-local buffer(s) from thread: cb-io-1-1
SEVERE: The web application [/mauricio] appears to have started a thread named [threadDeathWatcher-4-1] but has failed to stop it. This is very likely to create a memory leak.
Note: The improvements have been partially backported to release11 branch for inclusion in the upcoming 2.1.5 release, but improvements for RxJava and Netty threads couldn’t be backported, so it’s mainly clarity of code, integration tests and logs that have been backported.
Any specific reason of why Netty threads couldn’t be backported?
Thanks,
Mauricio