Fluentd Log Ordering Issue in Multi-Worker

[mikankun1003]

I'm using Fluentd v1.16.1 in a multi-worker environment:

• Forwarders use in_tail to read logs and out_forward to transfer them.
• The Aggregator uses in_forward to receive logs and out_file (with Chunk_keys Tag file Buffer) to write them to files.

We're running a single Aggregator with multiple workers. When the Aggregator encounters an issue, each Forwarder buffers Chunks locally and transfers these local Chunks simultaneously upon recovery.

This leads to the Aggregator temporarily receiving a large volume of logs. As each Worker processes these asynchronously, it can result in logs being written out of order.

Is there a way to guarantee the order of logs?

Note: It would be ideal if we could sort the log records before writing them to the file.