* Do not stop retrying when an exception is encountered. Save & propagate last exception if retry count is exceeded. * Add one more log message to help with debugging * Limit schema registry heap to attempt to control OOMs