Description
If smartmontools are configured correctly, i.e. through
files /etc/default/smartmontools and
/etc/smartd.conf they automatically notify on critical
events via system-local e-mail and also keep track of the state of the
drives under /var/lib/smartmontools.
Apart from that, not much help in attaching the data to monitoring systems is offered. Also, thinking of the SMART data as a “pre-failure” warning mechanism, it may make sense to check the state of the drives on a regular basis. If such a check is not supported by a tool, it can be forgotten rather easily.
This package contains scripts that are intended to supplement smartmontools’ capabilities by formatting the last days’ SMART values in a text-based table. The table shows the values of the attributes over multiple days, allowing the user to identify any noteworthy or unexpected changes.
Additionally, a script log_nvme_state.pl is supplied
together with a Systemd Timer file to collect similar data from NVMe
drives.
To build this package use ant package.
These scripts are not designed for standalone operation. Rather, the intended usage is to integrate them in a place where one regularly views the data such as e.g. logwatch, monit or similar tools.
smart_csv_to_table.pl
- Tracking the State of HDDs and SSDs
For typical SATA drives, smartd automatically stores
information about all of the SMART attributes in CSV files under
/var/lib/smartmontools.
E.g. for a typical SSD this might look as follows:
$ head -n 5 /var/lib/smartmontools/attrlog.Samsung_SSD_850_EVO_2TB-*.ata.csv
2025-05-13 19:23:07; 5;100;0; 9;97;10736; 12;98;1825; [...]
2025-05-13 19:53:07; 5;100;0; 9;97;10736; 12;98;1825; [...]
2025-05-13 20:23:07; 5;100;0; 9;97;10737; 12;98;1825; [...]
2025-05-13 20:53:07; 5;100;0; 9;97;10737; 12;98;1825; [...]
2025-05-13 21:23:07; 5;100;0; 9;97;10738; 12;98;1825; [...]
$ tail -n 5 /var/lib/smartmontools/attrlog.Samsung_SSD_850_EVO_2TB-*.ata.csv
2025-08-31 17:13:35; 5;100;0; 9;97;11309; 12;98;1924; [...]
2025-08-31 17:43:35; 5;100;0; 9;97;11309; 12;98;1924; [...]
2025-08-31 18:13:35; 5;100;0; 9;97;11310; 12;98;1924; [...]
2025-08-31 18:43:35; 5;100;0; 9;97;11310; 12;98;1924; [...]
2025-08-31 19:13:35; 5;100;0; 9;97;11311; 12;98;1924; [...]The format is as follows: Each line contains the point in time followed by the values of all attributes at that time in the following format:
Attribute ID;Normalized Value;Raw Value;This could be analyzed by a spreadsheet, but the tools from this
package offer a script to analyze this data. It is called
smart_csv_to_table.pl and processes the most recent entries
(for the current month) of such a given table joining them with a
human-readable information about the meaning of the attribute IDs.
In order to generate such an information about the meaning of the
attribute IDs, smartctl can be used to place a file
monitoring.NAME.txt next to the log file
attrlog.NAME.csv under /var/lib/smartmontools,
e.g. by using the following command line
smartctl -A /dev/disk/by-id/ata-Samsung_SSD_850_EVO_2TB_S2RMNX0H900402N > /var/lib/smartmontools/monitoring.Samsung_SSD_850_EVO_2TB-S2RMNX0H900402N.ata.txtThe content of
monitoring.Samsung_SSD_850_EVO_2TB-S2RMNX0H900402N.ata.txt
looks e.g. like this:
smartctl 7.4 2023-08-01 r5530 [x86_64-linux-6.1.0-38-amd64] (local build)
Copyright (C) 2002-23, Bruce Allen, Christian Franke, www.smartmontools.org
=== START OF READ SMART DATA SECTION ===
SMART Attributes Data Structure revision number: 1
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH [...] RAW_VALUE
5 Reallocated_Sector_Ct 0x0033 100 100 010 [...] 0
9 Power_On_Hours 0x0032 097 097 000 [...] 11311
12 Power_Cycle_Count 0x0032 098 098 000 [...] 1924
177 Wear_Leveling_Count 0x0013 098 098 000 [...] 33
179 Used_Rsvd_Blk_Cnt_Tot 0x0013 100 100 010 [...] 0
181 Program_Fail_Cnt_Total 0x0032 100 100 010 [...] 0
182 Erase_Fail_Count_Total 0x0032 100 100 010 [...] 0
183 Runtime_Bad_Block 0x0013 100 100 010 [...] 0
187 Uncorrectable_Error_Cnt 0x0032 100 100 000 [...] 0
190 Airflow_Temperature_Cel 0x0032 065 055 000 [...] 35
195 ECC_Error_Rate 0x001a 200 200 000 [...] 0
199 CRC_Error_Count 0x003e 100 100 000 [...] 0
235 POR_Recovery_Count 0x0012 099 099 000 [...] 22
241 Total_LBAs_Written 0x0032 099 099 000 [...] 63173123890As you can see, the ID 5 corresponds to Reallocated_Sector_Ct, 9 is Power_On_Hours etc. The script combines this information into a neat table:
$ smart_csv_to_table.pl /var/lib/smartmontools/attrlog.Samsung_SSD_850_EVO_2TB-*.ata.csv
Attribute 25 26 27 28 29 30 31
9 Power On Hours 11293 11294 11298 11299 11300 11303 11311
12 Power Cycle Count 1918 1919 1920 1921 1922 1923 1924
177 Wear Leveling Count 98:33 98:33 98:33 98:33 98:33 98:33 98:33
190 Airflow Temperature Cel 76:24 67:33 73:27 74:26 73:27 73:27 65:35
235 POR Recovery Count 99:22 99:22 99:22 99:22 99:22 99:22 99:22
241 Total LBAs Written 631~434 631~042 631~154 631~898 631~938 631~730 631~594The first row consists of the number of the days (e.g. here 2025-08-25 .. 2025-08-31) and the first column contains a description for the ID and attribute name with the underscores replaced by spaces for readability.
Not all attributes are reproduced: The ones where the normalized value is 200 or 100 for all columns are skipped because they commonly indicate “perfect condition” variables which are only rarely of interest.
Finally, there are three styles by which the numbers are displayed compactly:
- If space permits e.g. because the numbers are very short, a colon-separated information is printed in the form NORMALIZED:RAW to show both values.
- If the number is too long for that type of display, only the raw value is shown.
- If the number is too long to fit the column width, the first and the
last three digits are shown and separated by a tilde
(
~).
log_nvme_state.pl
- Tracking the States of NVMe Drives over Time
For a long time, smartmontools didn’t support the
logging of attributes of NVMe drives, cf. https://www.smartmontools.org/ticket/1190.
If you have access to a reasonably recent smartmontools version, it may make sense to prefer the logging implemented by smartmontools over the external scripts presented in the following.
The script log_nvme_state.pl provides a way to query the
state of all NVMe drives in the system (using the nvme
utility) and store the associated attributes in a SQLITE database.
SYSTEMD Support Files
In order to collect the required data, log_nvme_state.pl
needs to be invoked periodically. The systemd unit files
(masysma-log-nvme-state.service and
masysma-log-nvme-state.timer) delivered as part of this
package enable this, causing the attributes to be collected once per
hour if the timer is enabled.
smart_sqlite_to_table.pl
- Formatting the NVMe Information as a Table
To visualize the attributes stored by log_nvme_state.ply
in a similar manner to smart_csv_to_table.pl, the script
smart_sqlite_to_table.pl may be used e.g. as follows:
$ smart_sqlite_to_table.pl /var/lib/smartmontools/attrlog.INTEL_SSDPE2KX040T8-*.nvme.sqlite
INTEL SSDPE2KX040T8 at /dev/nvme1n1
Attribute 28 29 30 31 01 02 03
avail spare 100 100 100 100 100 100 100
controller busy time 246 246 246 247 247 247 247
data units read [TiB] 100.95 100.95 100.96 100.98 101.01 101.02 101.02
data units written [TiB] 25.71 25.71 25.74 25.85 25.86 25.87 25.89
host read commands 1.6e09 1.6e09 1.6e09 1.6e09 1.6e09 1.6e09 1.6e09
host write commands 1.0e09 1.0e09 1.0e09 1.0e09 1.0e09 1.0e09 1.0e09
power cycles 1986 1987 1988 1989 1990 1991 1992
power on hours 11034 11035 11038 11050 11051 11053 11056
spare thresh 10 10 10 10 10 10 10
temperature 317 316 317 318 317 317 317
unsafe shutdowns 28 28 28 28 28 28 28Since there are no “normalized” values to deal with for NVMe drives,
the data is displayed as raw value by default. For large numbers,
scientific notation is used. For values with known units
(e.g. data_units_read), a meaningful unit (TiB) is
used.
Previously Used: A PHP-based Evaluation Script
Before attempting to integrate the SMART value report with other
tools using a common text-based format, a similar visualization of
attribute values was created using a PHP script (cf.
old_php_smartd_auswertung/php_smartd_auswertung_quickm.php).
For using this script, it was necessary to establish an XML-based
configuration (php_smartd_auswertung_conf.xml) that would
contain the meanings of the attribute IDs and also give a hint as to
whether a change of the attribute value was expected (e.g. Power on
Hours) or not (e.g. Reallocated Sector Count). The script would then
paint the cells in the table in different colors, using a visually
striking magenta to picture unexpectedly changing values and using
pastel colors to show good or “change expected” values.
In the example, the “POR Recovery Count” attribute unexpectedly increased from 19 to 22 on 2025-08-24 hence the magenta marker.
The PHP script is old and not very convenient to use. Still, it is included here because it still makes sense to keep the colorized tabular representation in mind as a useful way to plot SMART values. It might be an inspiration for future development on SMART attribute analysis and monitoring tools.
Future Directions
Currently, Ma_Sys.ma systems do not ever run the SMART self-tests. This may be something worth enabling (at least on HDDs) in the future.