Ad Widget

Collapse

Getting 'mount <local_mount> missing' alerts.

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • osk_dt
    Junior Member
    • Jan 2025
    • 8

    #1

    Getting 'mount <local_mount> missing' alerts.

    We have some servers with NFS or CIFS mounts. We started noticing that when we have issues with the NFS mount, we sometimes get alerts indicating all our local mounts are missing. For example, if the NFS mount hangs (i.e. can't connect to the NFS server) we will start getting alerts indicating all local mounts are missing (i.e. /usr is missing).

    I don't know how the mount checking works in Zabbix, but the behavior mimics the behavior of running a 'df' on all file systems. Since NFS is hanging, you don't get any data for any of the file systems. Based on the behavior, it looks like you're missing all your local mounts. Is Zabbix doing something similar? Is it trying to collect data on all mounts, including the NFS mount, then failing to get any data because the NFS mount hangs? If so, is there a way to tell Zabbix to exclude the NFS mounts so it can get the data for the other mounts?
  • cyber
    Senior Member
    Zabbix Certified SpecialistZabbix Certified Professional
    • Dec 2006
    • 4807

    #2
    There is too little info here... What template are you using, what items(keys/discovery ? You say, you don't know how mount checking is working.. but neither do we, if you don't enlighten us with details..

    Comment

    • osk_dt
      Junior Member
      • Jan 2025
      • 8

      #3
      I'm not sure which check is indicating the missing file systems. We are using the default "Linux by Zabbix agent active" template. I tried getting a dump of active checks and I think one of the following ones might be the check indicating the missing file system. Is there a way to find out which specific check is seeing the file systems as missing?
      Thanks,

      "vfs.file.size[/etc/passwd]"
      "vfs.file.time[/etc/passwd,modify]"
      "vfs.file.exists[/etc/passwd]"
      "vfs.file.contents[/etc/passwd]"
      "vfs.file.regexp[/etc/passwd,root]"
      "vfs.file.regmatch[/etc/passwd,root]"
      "vfs.file.md5sum[/etc/passwd]"
      "vfs.file.cksum[/etc/passwd]"
      "vfs.file.owner[/etc/passwd]"
      "vfs.file.permissions[/etc/passwd]"
      "vfs.file.get[/etc/passwd]"
      "vfs.dir.size[/var/log]"
      "vfs.dir.count[/var/log]"
      "vfs.dir.get[/var/log]"
      "vfs.fs.size[/,free]"
      "vfs.fs.inode[/,free]"
      "vfs.fs.discovery"
      "vfs.fs.get"
      "vfs.dev.write[sda,operations]"

      Comment

      • markfree
        Senior Member
        • Apr 2019
        • 868

        #4
        I think you may be mixing your Zabbix alerts with error messages from your system.
        If you're getting alerts from Zabbix, there must be some events coming from your monitored host.
        You should follow the event to see what kind of trigger and item is being evaluated.

        Then, rewrite your question with examples and pictures, if possible.

        Comment

        • osk_dt
          Junior Member
          • Jan 2025
          • 8

          #5
          I think this is the information. When I looked at the problem, it shows the "Mount /data is missing" is coming from a space check for /data. Mark, this what you're talking about? I assume the check is 'Get data.' If that is the case, how is it collecting that data? Is it simply doing a 'df' on all file systems and the reason it's listing as missing is because it times out? I do believe the "mount missing" issue is related to the NFS mounts hanging.



          ​​

          Comment

          • cyber
            Senior Member
            Zabbix Certified SpecialistZabbix Certified Professional
            • Dec 2006
            • 4807

            #6
            There is no such trigger in default template as "Mount XXXXX is missing" ..


            None of the items you listed is related to /data muntpoint... Even if vfs.fs.discovery might find it, there is no trigger prototype for "... is missing"...

            Comment

            • osk_dt
              Junior Member
              • Jan 2025
              • 8

              #7
              I don't know if it makes a difference or not, but it's a trigger prototype. We're using Zabbix 7.0 LTS and the trigger is under "Linux by Zabbix agent active." I'm fairly certain this isn't something we added.

              Click image for larger version

Name:	image.png
Views:	74
Size:	36.5 KB
ID:	502754

              Comment

              • markfree
                Senior Member
                • Apr 2019
                • 868

                #8
                As Cyber said, the trigger you display is not present on the default "Linux by Zabbix Agent active" template.
                It appears that your template has been customized.

                The trigger expression is simple: it checks if the latest value for the total filesystem size is zero, or if no data has been received in the last 30 minutes, as the name suggests.
                The "Get filesystems" item is the master item that actually collects all data. You can "test" the item to see all the raw data retrieved.

                From what I have gathered so far, this trigger should only evaluate the discovered filesystem. If you are receiving alerts for the "local" FS, check your hosts' mounts.

                Comment

                Working...