next retry interval at XYZwhich can help you diagnose this issue.
retry_max_intervalto the output plugin configuration section to cap the max amount of time to wait until attempting a retry
flush_intervalis low enough that you are continuously flushing the buffer as you are reading data. For example if you are reading 10,000 events / second make sure you are not flushing data every hour otherwise your buffer can quickly fill up
flush_thread_count. if you have excessive messages per second and Fluentd is failing to keep adjusting these two settings will increase Fluentd's resource utilization but hopefully allow the application to keep up with the required throughput
total_limit_sizefile buffers are also much larger (64GB file vs. 512 MB memory)
total_limit_sizeas well as changing the maximum size of chunks
buffer_pathand has not been flushed