cc-metric-collector/collectors/diskstatMetric.md

diskstat collector

  "diskstat": {
    "exclude_metrics": [
      "disk_total"
    ],
    "exclude_devices": [
      "slurm-tmpfs"
    ],
    "exclude_mountpoints": [
      "/tmp"
    ],
    "mountpoint_as_stype": true,
    "use_include_config": false,
    "include_devices": [
      "/dev/sda3"
    ],
    "include_mountpoints" : [
      "/home"
    ]
  }

The diskstat collector reads data from /proc/self/mounts and outputs a handful node metrics with stype=filesystem,stype-id=<mountdevice>. If a metric is not required, it can be excluded from forwarding it to the sink.

For sending the mountpoint instead of the mountdevice in the stype-id, use mountpoint_as_stype.

There are two ways to specify for which devices or mountpoints the collector generates metrics. It's "either ...or".

• Excluding devices and mount points using exclude_devices and exclude_mountpoints. All devices (*) will be read that are not explicitly excluded
• Include devices and mount points by setting use_include_config:true and using include_devices and include_mountpoints.

(*) File systems where the mount device (first column in /proc/self/mounts) contains loop are always excluded. Filesystems where the mount point (second column in /proc/self/mounts) contains boot are also always excluded.

Metrics per filesystem (with stype=filesystem tag and stype-id based on the configuration):

• disk_total (unit GBytes)
• disk_free (unit GBytes)

Global metrics (with stype=filesystem tag and stype-id pointing to the max. used filesystem device or mount point based on the configuration):

• part_max_used (unit percent)