Uploaded image for project: 'Observium'
  1. Observium
  2. OBS-1826

Add GPU and memory usage for unix-agent

Details

    • Improvement
    • Resolution: Fixed
    • Minor
    • None
    • Professional Edition
    • Poller

    Description

      Hi,

      The unix-agent nvidi-smi script already retrieve all required info but includes/polling/unix-agent/nvidia.inc.php only processes the power and temperature data.

      Could GPU usage and memory usage be added ?

      I have attached a debug poller of a node with the unix-agent and 4 Nvidia telsa cards.

      Thanks.

      Attachments

        Activity

          [OBS-1826] Add GPU and memory usage for unix-agent
          landy Mike Stupalov added a comment -

          Was added long time ago, with some fixes in r12202.

          landy Mike Stupalov added a comment - Was added long time ago, with some fixes in r12202.
          lobaille Louis Bailleul added a comment -

          Hi,

          Bumping this one as I realized that you can get the actual values for memory (total,used,free).
          That might make it easier to add support without making changes.

          I have attached the modified nvidia-smi script.

          sample output :

          nvidia-smi --format=csv --query-gpu=index,gpu_name,driver_version,temperature.gpu,fan.speed,power.draw,pstate,utilization.gpu,memory.total,memory.free,memory.used

          index, name, driver_version, temperature.gpu, fan.speed [%], power.draw [W], pstate, utilization.gpu [%], memory.total [MiB], memory.free [MiB], memory.used [MiB]

          0, Quadro M4000, 352.93, 37, 0 %, 41.35 W, P0, 0 %, 8191 MiB, 8172 MiB, 19 MiB

          lobaille Louis Bailleul added a comment - Hi, Bumping this one as I realized that you can get the actual values for memory (total,used,free). That might make it easier to add support without making changes. I have attached the modified nvidia-smi script. sample output : nvidia-smi --format=csv --query-gpu=index,gpu_name,driver_version,temperature.gpu,fan.speed,power.draw,pstate,utilization.gpu,memory.total,memory.free,memory.used index, name, driver_version, temperature.gpu, fan.speed [%] , power.draw [W] , pstate, utilization.gpu [%] , memory.total [MiB] , memory.free [MiB] , memory.used [MiB] 0, Quadro M4000, 352.93, 37, 0 %, 41.35 W, P0, 0 %, 8191 MiB, 8172 MiB, 19 MiB
          adama Adam Armstrong added a comment -

          Processors, maybe. Memory, not without significant changes, as it's only reporting a percentage, and not actual numbers.

          adama Adam Armstrong added a comment - Processors, maybe. Memory, not without significant changes, as it's only reporting a percentage, and not actual numbers.

          People

            landy Mike Stupalov
            lobaille Louis Bailleul
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: