Filebeat kubernetes reuse inode

[mafr (Marcin)]

Hello,

i have got a kubernetes enviroment (v1.19.2) with some deployments.

Filebeat autodiscover properly "detects" log files, but after first log collect , filebeat are not harvest log files again (maybe after some time?)
Probably it is due to inode reuse on docker log file.
I'm using ext4 filesystem

Docker daemon.json

 max-files = 1, 
 max-size = 100M

Filebeat.yml properties:

    filebeat.autodiscover:
      providers:
        - type: kubernetes
          node: ${HOSTNAME}
          ignore_older: 1s
          scan_frequency : 1s
          clean_inactive: 3s

How I can deal with inode reuse?
Filebeat can deal with 40k log lines per second?
Docker can reach max file size in 10s
As output i'm using logstash

ELKF 7.12.0