WD NAS App Builder
Node Exporter 1.12.1-1
The NAS's hardware, system and WD storage metrics for Prometheus and Grafana
monitoring
Download
| NAS | Size | SHA256 | |
|---|---|---|---|
| My Cloud PR4100 | 12 MB | Download | ef0aeb86a118c294… |
| My Cloud PR2100 | 12 MB | Download | 236b634a5351f4ce… |
| My Cloud DL4100 | 12 MB | Download | 868223e8475dcd19… |
| My Cloud DL2100 | 12 MB | Download | 08e7999f047908cf… |
| My Cloud EX4100 | 11 MB | Download | a04182ab3636c02c… |
| My Cloud EX2100 | 11 MB | Download | c469773eea134e9e… |
| My Cloud EX2 Ultra | 11 MB | Download | 4fbf2e82614a61cc… |
| My Cloud Mirror (Gen 2) | 11 MB | Download | 3d6b70147b51840f… |
| My Cloud (Gen 2) | 11 MB | Download | 069343d6a1e60a0e… |
| WD Cloud | 11 MB | Download | 3c8ab2b42c875c37… |
Easier: install it from the Store, which also updates it.
Node Exporter is Prometheus's official exporter for a machine's hardware and system metrics (CPU, memory, load, disk I/O, filesystems, network, temperatures, software RAID…). Prometheus collects them from it, and Grafana shows them.
- The metrics are at
http://<nas-ip>:9100/metrics(the port can be changed on the management page). - The app adds WD's own view of the storage, which Linux alone doesn't show, as metrics starting with
wd_(see below). - Nothing is written to the disks while it runs: its files and log are kept in RAM (
/tmp), and WD's records are read rather than the disks themselves, so sleeping disks stay asleep. - Its settings (port, password) are kept next to the app, e.g.
/mnt/HD/HD_a2/Nas_Prog/_node_exporter, and survive reinstalls. - Node Exporter doesn't update itself: updates come with new versions of the app, and the NAS shows an alert (and sends an email, if set up) when a new release is out, checked weekly.
WD storage metrics
A small script reads WD's records of the disks, RAID arrays and volumes every 30 seconds (the same ones the dashboard's Storage pages use, via WD's xmldbc), and Node Exporter serves them with its own metrics (through its textfile collector):
| Metric | Labels | Value |
|---|---|---|
wd_health | Overall storage health: 0 healthy, 1 degraded (e.g. a disk too hot, a RAID rebuilding), 2 critical (e.g. a failed disk, a degraded RAID, a volume not mounted) | |
wd_info | model, firmware | Always 1 |
wd_disk_info | disk, bay, vendor, model, firmware | Always 1 |
wd_disk_temperature_celsius | disk, bay | The temperature WD last read |
wd_disk_healthy, wd_disk_failed, wd_disk_over_temperature, wd_disk_sleeping | disk, bay | 1 or 0 |
wd_disk_smart_passed | disk, bay | Whether the last SMART test WD ran passed (only once one has run) |
wd_disk_size_bytes | disk, bay | |
wd_raid_info | raid, level, system | Always 1 (system="1" is WD's own small system array, md0) |
wd_raid_state | raid, state | Always 1: the state is the label (e.g. clean, degraded, recovering) |
wd_raid_disks | raid, state | Disk counts: total, active, working, spare, failed, rebuilding |
wd_raid_dirty | raid | 1 or 0 |
wd_raid_size_bytes | raid | |
wd_volume_info | volume, label, mountpoint, raid, raid_level | Always 1 |
wd_volume_mounted, wd_volume_encrypted, wd_volume_unlocked | volume | 1 or 0 |
wd_volume_size_bytes, wd_volume_used_bytes | volume | |
wd_collector_success | Whether WD's records could all be read |
Disk serial numbers are left out. For alerts, e.g. wd_health > 0, wd_disk_temperature_celsius > 55 or wd_raid_disks{state="failed"} > 0.
SMART metrics
WD's records only say whether a disk has passed or failed. Every 5 minutes the script also reads each disk's own SMART data with smartctl (smartmontools 7.5, built static by static/node_exporter for the Intel models - the ARM models use the NAS's own smartctl, if it has one), with -n standby: a sleeping disk isn't woken, and its last reading is kept until it wakes.
| Metric | Labels | Value |
|---|---|---|
wd_smart_healthy | disk, bay | The disk's own health assessment: 1 passed, 0 failed |
wd_smart_attribute_raw | disk, bay, id, attribute | Each attribute's raw value (its first number - e.g. sectors, hours, degrees) |
wd_smart_attribute_value, wd_smart_attribute_threshold | the same | The normalised value (higher is better), and the threshold at or below which it counts as failing |
wd_smart_attribute_failing | the same | 1 if the attribute is failing now or has failed before |
wd_smart_last_read_timestamp_seconds | disk, bay | When the values were last read |
wd_smart_asleep | disk, bay | 1 if the disk was asleep at the last attempt (so the values are from before) |
wd_smart_success | Whether SMART values could be read for at least one disk |
The attributes that warn of a failing disk are 5 (Reallocated_Sector_Ct), 197 (Current_Pending_Sector) and 198 (Offline_Uncorrectable) - any above 0 deserves attention, and a rising count means trouble - and 199 (UDMA_CRC_Error_Count: cable or connection errors). For alerts, e.g. wd_smart_attribute_raw{id=~"5|197|198"} > 0 or wd_smart_healthy == 0. The Grafana app's WD NAS dashboard shows them per disk, and Node Exporter's page shows each disk's SMART health and problem sectors.
Management page
Node Exporter's page in the NAS dashboard (Apps → Node Exporter) shows whether it's running, its version and metrics address, and the storage as WD sees it (disks with temperatures and SMART health, RAID arrays, volumes and overall health). It changes the port, can require a password (see below), gives a prometheus.yml snippet to copy, and shows Node Exporter's latest warnings. Only NAS admins can use the page (see Management pages).
Password
Like Node Exporter on other machines, the metrics can be read by anyone on the local network by default (they show the NAS's activity, not its files or settings). The management page can require a password: tick the box, enter a password (or leave it empty to generate one) and save. The user name is prometheus, and the password can be shown, copied or changed on the page afterwards. It's saved, readable by root only, in Nas_Prog/_node_exporter/password, and Node Exporter is given only its hash (in web-config.yml). Add the basic_auth lines from the page's snippet to Prometheus.