4 Replies Latest reply on Sep 10, 2015 12:00 AM by Garland Smith
      • 1. Re: Km Unix FileSystem
        Garland Smith

        PKM for Unix uses an internal binary, statvfs, on most systems.

         

        statvfs is used because of potential hanging when attempting to

         

        access nfs-mounted filesystems.

         

         

         

        I provided some PSL code in a previous post that emulates the logic

         

        used by PKM for UNIX for filesystem monitoring.  If the following

         

        link doesn't get you there, look for "statvfs" on devcon and you should

         

        be able to find it.

         

         

         

        https://communities.bmc.com/communities/message/160668#160668

         

         

         

        Regards,

         

        Garland Smith

        • 2. Re: Km Unix FileSystem

          If you are using the bgsagent/collector then the value comes straight from the kernel. So there is no command.

           

          Jon

           

          Sent from my iPhone

          1 of 1 people found this helpful
          • 3. Re: Km Unix FileSystem
            Jon Trotter

            Is the same command statvfs used also to collect the information about the filesystem, such as

            FSAvailableSpace, FSAvailableSpaceMB, FSCapacity, etc.? Having an issue with data collection, but the agent config looks correct.  Checking with the admins to ensure the agent's permissions have not changed.

             

            Based on what Coop is saying, if the information is coming straight from the kernel, would patches, updates and the like cause an issue?  I'm assuming permissions would still be a concern?

            • 4. Re: Km Unix FileSystem
              Garland Smith

              statvfs is a binary that was implemented in order emulate "df" without the possibility of hanging if nfs server hangs.  The FILESYSTEM KM uses statvfs on most OS versions.  It may use a different methodology on Linux, I don't remember.  However, it has never used information straight from the kernel.  When we started using the Perform data collector, we split out the FILESYSTEM monitoring (I believe) because the Perform collector was launching df commands and therefore there was no inherent benefit to using Perform for collecting FILESYSTEM metrics.

               

              Garland Smith