Hi team, Just checking in — is the issue reported ...
# fluent-bit
a
Hi team, Just checking in — is the issue reported in #9021 considered fully resolved? We're still observing similar behavior in version 4.0.3, where files don't seem to be removed or stop being tracked after crossing the
Ignore_Older
threshold. Would appreciate any clarity on whether this is expected in 4.0.3?
l
do you have a repro? that as starting point would be useful
a
Yes, here's the relevant config snippet:
Copy code
pipeline:
  inputs:
    - name: tail
      refresh_interval: 5
      path: ${ENV_LOG_PATH}
      path_key: path
      tag: tail
      ignore_older: 3h
      rotate_wait: 120
      read_from_head: true
      storage.type: filesystem
      storage.pause_on_chunks_overlimit: true
      db: /mnt/logs/fluentbit/fluentbitpos.db
      buffer_max_size: 20M
      mem_buf_limit: 200MB
      threaded: on
I’m monitoring the file using
lsof -p <PID of fluentbit>
and it shows:
Copy code
[root@test-primary-0 svcusr]# ls -lart /mnt/data/usr/xyz/abc/db/jobs/logs/25655089.log
-rw------- 1 svcusr svcusr 2048 Aug 21 22:11 /mnt/data/usr/xyz/abc/db/jobs/logs/25655089.log
Current time on the system:
Copy code
[root@test-primary-0 svcusr]# date
Fri Aug 22 02:53:56 UTC 2025
Deployed on AWS, and tailing from EBS as well as EFS.. and EFS tends to have limits with number of files that can stay open at a time.
Copy code
<https://docs.aws.amazon.com/efs/latest/ug/troubleshooting-efs-fileop-errors.html#diskquotaerror>
Let me know if you need more detail.