Fluentd misses log rotation

[BalogAbel]

We are using Fluentd to read logs from pods in our OpenShift clusters, and forwarding these logs to Kafka. We are using Prometheus to monitor our Fluentd instances and we've found that sometimes Fluentd misses a log rotation. The skipped logs are not processed later, they are lost.
Fluentd version: 1.14.0

Fluentd in_tail config

<source>
  @type tail
  @id myproject_logs
  path /var/log/pods/myproject_*/*/*.log,/var/log/pods/openshift-ingress_router-myproject-*/*/*.log,
  follow_inodes true
  pos_file /var/log/fluent/fluentd-myproject-containers.log.pos
  tag "business_pods.*"
  read_from_head true
  rotate_wait 1
  enable_stat_watcher false
  <parse>
    @type tsv
    keys time,stream,logtag,log
    time_format '%Y-%m-%dT%H:%M:%S.%N%:z'
    time_key time
    keep_time_key true
    delimiter ' '
  </parse>
</source>

Logs are attached.
logs.log
As far as I can see, the issue is somewhere during the log rotation, as the logs disappers when the log rotation occurs (2022-07-29 11:17:01) and continue reading at 2022-07-29 11:33:01.

Input metrics:
The number of read logs for the problematic pod (file):

Do you have any idea what can cause this issue?

Thanks in advance, Abel