Message of the Day (MOTD) with ZFS Enhancements

Collection of 'Message of the Day' scripts with ZFS Enhancements

Requirements

• update-motd
• figlet & lolcat (for 10-hostname)
• smartmontools & nvme-cli (for 36-diskstatus)

How do I set it up?

Copy the files you want in your MOTD to /etc/update-motd.d/. Make sure you have the PrintMotd option set to yes in your sshd config.

The duplicate files are different versions of the same, use either one of them. E.g. 30-zpool-simple will not print usage bars.

The script 36-diskstatus will grep either journalctl or syslog for smartd entries to read last self-test result. For non-NVMe devices, you have to enable smartd monitoring & run regular self-tests for it to display anything. The nvme client is required to get NVMe device errors and wear leveling.

If you use 50-fail2ban you should comment out the compress option in /etc/logrotate.d/fail2ban, so that the logs are not compressed and can be read by grep.

---

HDDTemp Removed

The hddtemp utility was once the primary way to monitor and gather drive temperature information. However HDDTemp project is considered dead and no longer maintained. It is no longer included in many distribution repositories. Support for hddtemp has been removed. This script will get device temperature from smartctl.

---

36-diskstatus has NVMe Support

Support for NVMe devices is provided in 36-diskstatus where it will include any device that starts with nvme- (in the /dev/disk/by-id directory):

#---[ You can updates these ]--------------------------------------------------
# Types of disks to look for. Used by awk to define disk by-id types to include
# IDE/SATA - you might add another for usb "usb-".
findthese="ata-|scsi-SATA_|nvme-"

Then any devices matching nvme-eui or nvme-nvme are removed from the list. This should leave just devices with the manufacture name such as Samsung, HP, Sabrent, WD, etc. If you have some odd name detected then you probably need to add another filter here.

# This is used by awk to remove unwanted disk devices which matched above.
ignorethese="ata-Samsung|nvme-(eui|nvme).*"

Lastly there is a filter which attempts to pretty up the device names for display, for NVMe devices it removes the nvme- prefix with the intent that the device name will now start with your manufacture name. It will remove references to _with_Heatsink popular with Samsung devices.

# This is used by sed to remove text from disk device names. This does not alter
# device selection like ones above.  This just helps to make disk device names
# nicer (smaller).
sed_filter="s/^scsi-SATA_//; s/^ata-//; s/Series_//; s/^nvme-//; s/_with_Heatsink//;"

Example Results:

• Samsung_SSD_980_1TB_316T is a Samsung device, model 980 with 1 TB capacity with last 4 digits of the serial number 316T.
• HP_SSD_EX950_1TB_3005 is a HP (Hewlett Packard) model EX950 with a 1 TB capacity with last 4 digits of serial number 3005.
• WDS100T3XHC-00SJG0_2993 shows how horrible WD (Western Digital) names their devices. This is a WD Black model SN750 but there is no way to know that by the device name. (You could use the filter above to rename if you like)

The idea behind this is should you have many devices and one reports FAILED, having part of the serial number will be helpful in identifying which device has a problem.

NVMe Device Temperature

If 36-diskstatus script detects a NVMe device, it will scrape the output of smartctl by looking for Temperature: and parsing the value such as 38 Celsius.

disk status:
  Samsung_SSD_980_1TB_316T (nvme0n1):   38C passed [97%]

A Celsius temperature from smartctl can be converted to Fahrenheit by setting variable convert_c_to_f to /bin/true (set to /bin/false for Celsius).

# hddtemp can already report F or C temperatures for SATA devices.  If this script
# grabs a temp from smartctl and it is reported in "C" set the following to 
# if you want that value converted to F.
convert_c_to_f=/bin/true

disk status:
  Samsung_SSD_980_1TB_316T (nvme0n1):   98F passed [97%]

NOTE: Some devices may go into sleep mode and report SLP instead of current temperature:

disk status:
  SSSTC_CVB-8D128-HP_F1PA (sda):  SLP | without error

---

Example showing a mix of SATA SSDs, SATA HDDs and NVMe devices

• Temperatures are converted to Fahrenheit
• Elevated temperatures are in yellow
• Wear level / life expectancy are expressed as a percentage from 100% and lowers towards 0%

---

Drive Testing Results

Without Error

When the Customized Message of the Day disk status: states a drive is without error it is parsing that information from the device's last test result:

disk status:
  WDC_WD120EFBX-68B0EN0_U62B (sda):  73F without error

• SATA SSD which support a wear level indicator will have the without error changed to passed (lowercase) to make room for the indicator. This is a life expectancy expressed as a percentage that starts at 100% and decreases towards 0%.

disk status:
  Samsung_SSD_860_EVO_1TB_043Z (sda):   78F  passed [87%]

PASSED or FAILED

When a SATA device test result is not found (either not supported, not performed, log purged, etc) then the smartctl self assessment status will be displayed which should be a simple PASSED or FAILED! value. It is still possible to show PASSED (uppercase) and have device issues. This is not an equivalent of without error.

This is expected with a new device that has not been tested yet, or an initial install where the scheduled test(s) have not run yet.

disk status:
  Crucial_MX500_1TB_SSD_3640 (sda):  105F  PASSED

NVMe device will use the nvme-cli to fetch error information from the device. If the last test, a result of zero is passed and a non-zero is error. An error indicates you need to investigate the device, start with something like sudo nvme error-log -e 1 /dev/nvme0n1 to return the last error log entry.

disk status:
  Samsung_SSD_980_1TB_316T (nvme0n1):  98F passed [97%]

Untested[]

When no test result can be found, and no self assessment value can be obtained from smartctl then the device will simply report untested[]. If any unexpected result is found that would be shown within the square brackets []. Empty brackets indicate absolutely nothing was found. This is expected with environments such as Virtual Box.

disk status:
  VBOX_HARDDISK_eb03 (sda):  ERR | untested[] 

Wear Level Indicator

SATA SSD and NVMe devices which report a detectable wear level indicator or a life expectancy indicator, will be expressed as a percentage that starts at 100% and decreases towards 0%. smartctl attribute names evaluated: Wear_Leveling_Count and Percent_Lifetime_Remain, NVMe attribute percentage_used.

The indicator will remain green until it drops below 30%, it will then be yellow to get your attention. Below 15% will turn red. The device should be replaced.

---

Originally Based on: https://github.com/yboetz/motd