zabbix-zfs

Zabbix template & script to monitor ZFS on Linux.

Single Python script that emits all information needed for discovery & data gathering in a single JSON. All items are defined as Dependent and extract relevant data using JSONPath queries.

{
  "vdevs": {
    "/dev/sda1": {
      "name": "/dev/sda1",
      "size": 0,
      "alloc": 0,
      "free": 0,
      "frag": 0,
      "usage": 0,
      "online": 1,
      "errors": {
        "read": 0,
        "write": 0,
        "cksum": 0
      }
    }
  },

  "pools": {
    "pool1": {
      "name": "pool1",
      "size": 11957188952064,
      "alloc": 4227267283968,
      "free": 7729921668096,
      "frag": 5,
      "usage": 35,
      "dedup": 1.0,
      "scrub": 0,
      "online": 1
    }
  },

  "datasets": {
    "pool1": {
      "name": "pool1",
      "avail": 5502826989670,
      "used": 3162194032538,
      "compress": 1.05,
      "referenced": 38300
    }
  },

  "arc": {
    "l1": {
      "size": 5398891600,
      "hitrate": 91.56803867935483,
      "free": 458843264
    },
    "l2": {
      "usage": 142278352384,
      "usage_actual": 137375770112,
      "hitrate": 94.00152633863829,
      "bytes_read": 241515501568,
      "bytes_written": 25932241920,
      "io_error": 0,
      "cksum_bad": 0
    }
  },

  "slab": 291337024
}

Features

• Low level discovery of:
  • Pools
  • Datasets
  • Devices
  • L2ARC - items will be created only if it's present
• Items:
  • Pools: health, disk usage, fragmentation, deduplication, scrub
  • Datasets: disk usage, compression ratio
  • Devices: health, disk usage, error counts, fragmentation
  • SLAB usage
  • L1 & L2 ARC stats
• Triggers:
  • Pool & device health
  • Pool & dataset disk usage
  • High fragmentation
  • ARC usage
  • L2ARC errors & low hit rate
  • Scrub
• Zabbix agent passive checks. Can be converted to active if needed.

Macros

• {$ZFS_DS_EXCLUDE} - Regexp to exclude the datasets during discovery
• {$ZFS_DS_INCLUDE} - Regexp to include the datasets during discovery
• {$ZFS_DS_USAGE_CRIT} - Dataset usage in % when crit alert is triggered
• {$ZFS_DS_USAGE_HIGH} - Dataset usage in % when high alert is triggered
• {$ZFS_POOL_EXCLUDE} - Regexp to exclude the pools during discovery
• {$ZFS_POOL_INCLUDE} - Regexp to include the pools during discovery
• {$ZFS_POOL_FRAG_HIGH} - Pool fragmentation in % when high alert is triggered
• {$ZFS_POOL_USAGE_CRIT} - Pool usage in % when crit alert is triggered
• {$ZFS_POOL_USAGE_HIGH} - Pool usage in % when high alert is triggered
• {$ZFS_VDEV_ERROR_THRESHOLD} - vdev and L2ARC error threshold when triggers fire
• {$ZFS_VDEV_EXCLUDE} - Regexp to exclude the vdevs during discovery
• {$ZFS_VDEV_INCLUDE} - Regexp to include the vdevs during discovery
• {$ZFS_ARC_META_THRESHOLD} - Alert when ARC meta usage % goes over this value
• {$ZFS_L2ARC_ERROR_THRESHOLD} - Error threshold for L2ARC errors
• {$ZFS_L2ARC_HITRATE_THRESHOLD} - Alert when L2ARC hit rate falls below this value

Requirements

• Tested on OpenZFS 2.x, maybe will work with 0.8.x
• Tested on Zabbix 5.0, but should work on 4.2+
• Python3

Installation

• Place zfs.conf in /etc/zabbix/zabbix_agentd.d
• Place zfs.py in /etc/zabbix/scripts You can put it into any other place, but then you'll have to adjust zfs.conf
• Restart zabbix-agentd
• Import template_zfs.xml
• You're good to go