we are having missing log issue in cloud watch when scaling more then 1 k pod .each pod have 1000 lines .

I have added below this has improved the situation but not completely..
Buffer_max_size 600MB
mem_buf_limit 750 MB
Skip_long_lines off
Refresh_interval 1
Rotate_wait 15
Inotify_watcher false
Storage.type filesystem
Buffer_chunk_size 100mb

And flush from 5 to 1 in service section

Need advice on how much more we can add on buffer size or any other configuration for fluent bit if we want to scale upto 20k pod

Thank you so much for your kind help ????

Tested above configuration for 5000 pod but still seeing log loss