we are running filebeat 6.8.22 with inputs as docker containers and it doesn't release file handlers for deleted files
filebeat.inputs:
- type: docker
close_inactive: 5m
close_removed: true
clean_removed: true
containers:
ids: "*"
path: /var/lib/docker/containers
stream: all
ignore_older: 36h
json.message_key: log
json.keys_under_root: false
json.ignore_decoding_error: true
scan_frequency: 60s
document_type: 'docker-logs'
processors:
- add_docker_metadata: ~
an example from ps
output
filebeat 27275 root 5r REG 202,1 100000031 127955205 /var/lib/docker/containers/2839281c157
8e88f15249ba1cad623f48831e7b1972d7ecbb36b1dfc1a0fba69/2839281c1578e88f15249ba1cad623f48831e7b1972d7ecbb36b1dfc1a0fba69-json.l
og.2 (deleted)
lsof -n|grep filebeat|grep delete|wc -l
285
this behavior causes extra disk space on EC2 nodes to be used and not properly cleaned up, and eventually, used space is hitting 100% percent and causing Out of disks space
errors, affecting other containers running at the same EC2 node
any ideas how to fix it? possible quick workaround can be periodic restart for filebeat
container, but it seems hacky to me
GitHub issue - filebeat not close delete filehandler · Issue #30472 · elastic/beats · GitHub