LogParallelismSlaveThread(XX):failed to push marker record (4304901)

Capture seems stuck and not processing any messages. Analyzing the ocap.log you find this message:
"LogParallelismSlaveThread(XX):failed to push marker record" where XX is the thread number capture is processing.

This message is "just" a notice. It means that the internal queue is full, whatever is pushing the messages into the internal queue (which is likely the log reader) will sleep and try again later. At that point the problems can start:

In 5.3.2.x all platforms Bug in microsleep function
In 5.3.3.x SUN platform Bug in nanosleep function
In 5.3.4.40 and lower SUN platform Deadlock bug

Bouncing capture helps in most cases as workaround but it's only a matter of time until capture will be stuck again. Upgrading to at least 5.3.4.50 is strongly advised!

All platforms, versions lower than 5.3.4.40, check with pthread if you hit the micro/nanosleep issue.
If yes, then bounce capture and upgrade as soon as possible to 5.3.4.50
5.3.4.x versions lower than 5.3.4.50 should upgrade to 5.3.4.50 (or higher)