Ad Widget

Collapse

Problem monitoring XFS filesystem

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • PD77
    Junior Member
    • Aug 2017
    • 5

    #1

    Problem monitoring XFS filesystem

    Hi,

    We have zabbix 3.0.2 installed in a centos 6.9 server.

    While monitoring an another centos client, Zabbix web interface displays incorrect disk usage details (total, used & free space) for partitions mounted as XFS but it works perfectly for ext4 filesystems.

    Any ideas why zabbix displays incorrect details for XFS in web interface ..? does any know how to fix this issue.

    Thanks so much your help in advance.

    Regards
    Padmaja.
  • kloczek
    Senior Member
    • Jun 2006
    • 1771

    #2
    So how did you found that something here is incorrect?
    http://uk.linkedin.com/pub/tomasz-k%...zko/6/940/430/
    https://kloczek.wordpress.com/
    zapish - Zabbix API SHell binding https://github.com/kloczek/zapish
    My zabbix templates https://github.com/kloczek/zabbix-templates

    Comment

    • PD77
      Junior Member
      • Aug 2017
      • 5

      #3
      when I run df -h command, /data partition (xfs) has total of 100G disk space but in the zabbix webinterface (graph), the respective total disk space of this XFS is showing as 50 G. Due to this issue, I am not getting any alerts, when disk space of /data genuinely exceeds its limits.


      Filesystem Size Used Avail Use% Mounted on
      /dev/mapper/vg_centos-lv_root
      26G 7.9G 17G 32% /
      tmpfs 3.8G 0 3.8G 0% /dev/shm
      /dev/sda1 477M 174M 279M 39% /boot
      /dev/sdb5 100G 56G 45G 56% /data


      I have attached screenshot of the zabbix graph. Hope you see it.

      Regards
      Padmaja.

      Comment

      • PD77
        Junior Member
        • Aug 2017
        • 5

        #4
        One more thing, I missed to add in my previous comment. when I run zabbix_agentd command against /data partition its gives fairly correct details.
        but in the web interface, it's populating entirely wrong disk space for /data partition (XFS).


        # zabbix_agentd -t vfs.fs.size[/data,total]
        vfs.fs.size[/data,total] [u|107320315904]

        # df -h | grep data
        /dev/sdb5 100G 56G 45G 56% /data

        Any ideas whats causing this issue.?

        Regards
        Padmaja.

        Comment

        • PD77
          Junior Member
          • Aug 2017
          • 5

          #5
          Please, did anybody experiencing this same problem? I desperately need to sort out this issue.

          Comment

          • quackduck
            Member
            • Feb 2025
            • 36

            #6
            Ran into a similiar problem today. Had to check the source code. XFS can report different metrics when it comes to disk usage for individual files, directories and disk space overall - the zabbix agent in simple terms doesn't do any distinnction and kind of doesn't know what it's doing. There's a difference beteen logical file sizes and disk allocation. XFS in a sense can compress trivial files (files containing only zeroes). It matters a lot, especially when monitoring.

            A workaround is to check the diskspace via "UserParameters" or whatever zabbix call remote commands executed by the agent.

            Comment

            Working...