We noticed leaking threads in the publishConfluence() method which adds 3 threads with each call to the thread pool and never give it back properly.

Over time we're ending up with thousands of unused threads and this kind of exceptions:

java.lang.OutOfMemoryError: unable to create new native thread

As a workaround, we were able to increase this limit, but we're still reaching it after some time without a restart.
 

Pipeline example code:

for (int i = 0; i < 1000; i++) {
    publishConfluence(
        editorList: [
            confluenceAfterToken(generator: confluenceText('.'), markerToken: 'my-marker')
        ],
        pageName: 'PageTest',
        siteName: 'confluence.example.de',
        spaceName: 'TestSpace'
    )
}

Result
I added a graph with the open threads of the master Jenkins process.
The first peak to ~3k was caused with the code above (adding 3k ".").
Afterwards we did a clean restart.
Then I built a version with this changed line in ConfluenceSession.java:

private final ListeningExecutorService executor = MoreExecutors.listeningDecorator(Executors.newFixedThreadPool(1));

-> Next run was only leaking 1k threads which somehow sounds like unclosed/enqueue is missing in the ConfluenceSession.