Unable to harvest log files with glob

Elastic Stack Beats
1

Hi Team,
I am trying to parse jenkins build logs using filebeat. There are several subdirectories that it needs to traverse to. I can also see in the filebeat log what all locations filebeat is checking. However when i create a new log file or add entry to existing log files, logstash is not harvesting those changes. Please help. If i use just a single path instead of 3 file path in my case, it seems to work.

My Filebeat config

[ashivaramakrishnan@ci001.atl1.turn.com filebeat]$ more conf.d/jenkins-ci.yml    
filebeat:
  inputs:
    - paths:
      - "/var/lib/jenkins/jobs/cd-pipeline-master*/builds/*/junitResult.xml"
      - "/var/lib/jenkins/jobs/cd-pipeline-release*/builds/*/junitResult.xml"
      - "/var/lib/jenkins/jobs/cd-pipeline-production*/builds/**/junitResult.xml"
      encoding: plain
      fields:
        timezone: PDT
        log_type: jenkins_junit
      close_renamed: true
      close_removed: true
      fields_under_root: true
      type: log
      ignore_older: 5m
      scan_frequency: 10s
      harvester_buffer_size: 16384

Filebeat log files::

Configured paths: [
/var/lib/jenkins/jobs/cd-pipeline-master*/builds/*/junitResult.xml 
/var/lib/jenkins/jobs/cd-pipeline-release*/builds/*/junitResult.xml 
/var/lib/jenkins/jobs/cd-pipeline-production*/builds/junitResult.xml 
/var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/*/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/*/*/*/*/junitResult.xml /var/lib/jenkins/jobs/cd-pipeline-production*/builds/*/*/*/*/*/*/*/*/junitResult.xml]

For example if try modifying a file in the below path. filebeat doesnt harvest the file.
eg::

/var/lib/jenkins/jobs/cd-pipeline-master-ucheetahclient-test/builds/0000/junitResult.xml

Thanks
Arvind

2

Hi @Arvind_Ks :slightly_smiling_face: I'm not fully sure if this is supported but I think that when using folders you need to use ** (double asterisk). So cd-pipeline-master* must be cd-pipeline-master** for example. Can you give it a try?

3

Hi Mario,
Thanks for replying back. Actually when i used "**" it was searching upto 8 sub-directories and filebeat was running slowly. Now i replaced that with a single "*" and it appears to be working fast.

Thanks
Arvind

4

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.