Discussion thread for official Zabbix SMART Disk monitoring

[INFinite]

@INFinite Please note that the template is designed to exclude some devices that do not support queries. There are two macros {$SMART.DISK.NAME.MATCHES} and {$SMART.DISK.NAME.NOT_MATCHES}
If you exclude /dev/sda, the errors will disappear.

But does smartctl --scan also output other devices on its list? You need to make sure of this.

I admit that I could have done something wrong, so I'll give you a screenshot. If everything is correct, then the result has not changed.

C:\Windows\system32>smartctl --scan
/dev/sda -d scsi # /dev/sda, SCSI device
/dev/sdb -d ata # /dev/sdb, ATA device
/dev/sdc -d sat # /dev/sdc [SAT], ATA device
/dev/csmi0,0 -d ata # /dev/csmi0,0, ATA device
/dev/csmi0,1 -d ata # /dev/csmi0,1, ATA device
/dev/nvme0 -d nvme # /dev/nvme0, NVMe device

There's a regular intel soft raid. In Agent version 6.4, it perfectly sees the disks behind this raid.
Here is the output from Agent 6.4:

C:\Windows\system32>"C:\Program Files\Zabbix Agent 2\zabbix_agent2.exe" -c "C:\P
rogram Files\Zabbix Agent 2\zabbix_agent2.conf" -t smart.disk.discovery
smart.disk.discovery                          [s|[{"{#NAME}":"csmi0,0","{#DISKTY
PE}":"ssd","{#MODEL}":"KINGSTON SV300S37A60G","{#SN}":"50026B7246018A2F","{#PATH
}":"/dev/csmi0,0","{#RAIDTYPE}":"","{#ATTRIBUTES}":"Raw_Read_Error_Rate Retired_
Block_Count Power_On_Hours_and_Msec Power_Cycle_Count Program_Fail_Count Erase_F
ail_Count Unexpect_Power_Loss_Ct Wear_Range_Delta Program_Fail_Count Erase_Fail_
Count Reported_Uncorrect Airflow_Temperature_Cel Temperature_Celsius ECC_Uncorr_
Error_Count Reallocated_Event_Count Unc_Soft_Read_Err_Rate Soft_ECC_Correct_Rate
 Life_Curve_Status SSD_Life_Left SandForce_Internal SandForce_Internal Lifetime_
Writes_GiB Lifetime_Reads_GiB"},{"{#NAME}":"csmi0,1","{#DISKTYPE}":"ssd","{#MODE
L}":"KINGSTON SV300S37A60G","{#SN}":"50026B7246018A2C","{#PATH}":"/dev/csmi0,1",
"{#RAIDTYPE}":"","{#ATTRIBUTES}":"Raw_Read_Error_Rate Retired_Block_Count Power_
On_Hours_and_Msec Power_Cycle_Count Program_Fail_Count Erase_Fail_Count Unexpect
_Power_Loss_Ct Wear_Range_Delta Program_Fail_Count Erase_Fail_Count Reported_Unc
orrect Airflow_Temperature_Cel Temperature_Celsius ECC_Uncorr_Error_Count Reallo
cated_Event_Count Unc_Soft_Read_Err_Rate Soft_ECC_Correct_Rate Life_Curve_Status
 SSD_Life_Left SandForce_Internal SandForce_Internal Lifetime_Writes_GiB Lifetim
e_Reads_GiB"},{"{#NAME}":"nvme0","{#DISKTYPE}":"nvme","{#MODEL}":"","{#SN}":"","
{#PATH}":"/dev/nvme0","{#RAIDTYPE}":"","{#ATTRIBUTES}":""},{"{#NAME}":"sdc","{#D
ISKTYPE}":"hdd","{#MODEL}":"ST1000LM024 HN-M101MBB","{#SN}":"S30CJ9CFC09569","{#
PATH}":"/dev/sdc","{#RAIDTYPE}":"","{#ATTRIBUTES}":"Raw_Read_Error_Rate Throughp
ut_Performance Spin_Up_Time Start_Stop_Count Reallocated_Sector_Ct Seek_Error_Ra
te Seek_Time_Performance Power_On_Hours Spin_Retry_Count Calibration_Retry_Count
 Power_Cycle_Count G-Sense_Error_Rate Power-Off_Retract_Count Temperature_Celsiu
s Hardware_ECC_Recovered Reallocated_Event_Count Current_Pending_Sector Offline_
Uncorrectable UDMA_CRC_Error_Count Multi_Zone_Error_Rate Load_Retry_Count Load_C
ycle_Count"}]]

Today, once again trying to find a solution on the Internet, I came across bug report ZBX-26098 with the same problem.

[PavelZ]

@INFinite Please note that the template is designed to exclude some devices that do not support queries. There are two macros {$SMART.DISK.NAME.MATCHES} and {$SMART.DISK.NAME.NOT_MATCHES}
If you exclude /dev/sda, the errors will disappear.

But does smartctl --scan also output other devices on its list? You need to make sure of this.

[PavelZ]

I noticed another obvious problem:
You show the disk /dev/sda Raid 1 Volume, i.e. consisting of two physical ones.
If smartctl returns one set of indicators. Obviously, it cannot combine two sets of data.
The device should not be called /dev/sda.
Usually for Windows, disks are polled separately with the -d driver specified.
There is some key or naming convention there.

"scsi_vendor": "Intel",\r\n "scsi_product": "Raid 1 Volume",\r\n

Try to find the correct command. It will be something like:
smartctl -a /dev/csmi0,1

I do not have this exact controller model and it is difficult for me to give a ready-made recipe. But I hope this information will help you.

[INFinite]

Technically, all historical versions are available. But it's not that easy to figure it out if you're not used to working with Git.
The situation is complicated by the presence of several branches. You need to switch branches and download along this path:
https://github.com/zabbix/zabbix/commits/master/templates/server/smart_agent2/template_module_smart_agent2.yaml

I installed the template version 7.0. Agent version 7.0.7, the result is the same:

Cannot fetch data.: got error executing worker pool: smartctl returned error: unknown error from smartctl.

[PavelZ]

Technically, all historical versions are available. But it's not that easy to figure it out if you're not used to working with Git.
The situation is complicated by the presence of several branches. You need to switch branches and download along this path:

https://github.com/zabbix/zabbix/commits/master/templates/server/smart_agent2/template_module_smart_agent2.yaml

The update interval was 10s.

But in the standard template this interval is 5 minutes. It is unknown what other errors such frequent polling may cause.

[INFinite]

Wait, you do understand that templates also need to be updated? The template version is also subject of control.
I'm not ready to remote debug this problem, but there are obvious considerations.
Try to achieve a situation where both the template is version 7.0 and the agent is version 7.0

Also, I suggest focusing on smartctl first and making the error go away

Template version 7.2-1.
Can you tell me where I can find the old versions of the template?

When replacing files from the archive with version 6.4, everything immediately worked.

exit status 4.

It rarely occurs, it just happened to be an example. I associate it with a frequent request for data. The update interval was 10s.

[PavelZ]

Wait, you do understand that templates also need to be updated? The template version is also subject of control.
I'm not ready to remote debug this problem, but there are obvious considerations.
Try to achieve a situation where both the template is version 7.0 and the agent is version 7.0

Also, I suggest focusing on smartctl first and making the error go away

exit status 4.

[INFinite]

Thanks for the hint in which direction to look.

[INFinite]

Hello.
I ran into this problem, when setting up SMART monitoring on windows(I tried it on different versions OS windows):

Cannot fetch data.: got error executing worker pool: failed to execute smartctl: "{\r\n \"json_format_version\": [\r\n 1,\r\n 0\r\n ],\r\n \"smartctl\":
{\r\n \"version\": [\r\n 7,\r\n 4\r\n ],\r\n \"pre_release\": false,\r\n \"svn_revision\": \"5530\",\r\n \"platform_info\": \"x86_64-w64-mingw32-2012r2\",
\r\n \"build_info\": \"(sf-7.4-1)\",\r\n \"argv\": [\r\n \"smartctl\",\r\n \"-a\",\r\n \"/dev/sda\",\r\n \"-j\"\r\n ],\r\n \"exit_status\": 4\r\n },\r\n \"local_time\":
{\r\n \"time_t\": 1740725274,\r\n \"asctime\": \"Fri Feb 28 09:47:54 2025 RTZ\"\r\n },\r\n \"device\": {\r\n \"name\": \"/dev/sda\",\r\n \"info_name\":
\"/dev/sda\",\r\n \"type\": \"scsi\",\r\n \"protocol\": \"SCSI\"\r\n },\r\n \"scsi_vendor\": \"Intel\",\r\n \"scsi_product\": \"Raid 1 Volume\",\r\n
\"scsi_model_name\": \"Intel Raid 1 Volume\",\r\n \"scsi_revision\": \"1.0.\",\r\n \"scsi_version\": \"SPC-3\",\r\n \"user_capacity\": {\r\n \"blocks\": 111357952,
\r\n \"bytes\": 57015271424\r\n },\r\n \"logical_block_size\": 512,\r\n \"scsi_lb_provisioning\": {\r\n \"name\": \"thin provisioned\",\r\n \"value\": 2,\r\n
\"management_enabled\": {\r\n \"name\": \"LBPME\",\r\n \"value\": -1\r\n },\r\n \"read_zeros\": {\r\n \"name\": \"LBPRZ\",\r\n \"value\": 0\r\n }\r\n },\r\n
\"rotation_rate\": 0,\r\n \"logical_unit_id\": \"0x61fa116d01000000001517ffff0aeb84\",\r\n \"device_type\": {\r\n \"scsi_terminology\":
\"Peripheral Device Type [PDT]\",\r\n \"scsi_value\": 0,\r\n \"name\": \"disk\"\r\n },\r\n \"smart_support\": {\r\n \"available\": false\r\n },\r\n
\"temperature\": {\r\n \"current\": 0,\r\n \"drive_trip\": 0\r\n },\r\n \"seagate_farm_log\": {\r\n \"supported\": false\r\n }\r\n}\r": exit status 4.

The data is coming, but it does not have the correct format.

Ubuntu 24
zabbix server 7.2.4
zabbix agent2 7.2.3
smartmontools 7.4
template is SMART by Zabbix agent 2

Help me please

Experimentally, I found out that the "SMART by Zabbix agent 2" template works with agent version 6.4.21 and possibly earlier. It does not work with versions 7.0.X or 7.2.X.

[INFinite]

INFinite ,
There are some changes in the new versions of the agent.
Would you like to try slightly older ones?

I propose a version 7.0.7

After these changes, it gives the following error:

Cannot fetch data.: got error executing worker pool: smartctl returned error: unknown error from smartctl.

[PavelZ]

INFinite ,
There are some changes in the new versions of the agent.
Would you like to try slightly older ones?

I propose a version 7.0.7

[PavelZ]

Does the method you describe reduce security? In what way?

Any of your actions affect security. Therefore, I will not take responsibility.

I am simply informing you that this is an alternative solution described immediately in the Readme as recommended.

[INFinite]

Hello.
I ran into this problem, when setting up SMART monitoring on windows(I tried it on different versions OS windows):

Cannot fetch data.: got error executing worker pool: failed to execute smartctl: "{\r\n \"json_format_version\": [\r\n 1,\r\n 0\r\n ],\r\n \"smartctl\":
{\r\n \"version\": [\r\n 7,\r\n 4\r\n ],\r\n \"pre_release\": false,\r\n \"svn_revision\": \"5530\",\r\n \"platform_info\": \"x86_64-w64-mingw32-2012r2\",
\r\n \"build_info\": \"(sf-7.4-1)\",\r\n \"argv\": [\r\n \"smartctl\",\r\n \"-a\",\r\n \"/dev/sda\",\r\n \"-j\"\r\n ],\r\n \"exit_status\": 4\r\n },\r\n \"local_time\":
{\r\n \"time_t\": 1740725274,\r\n \"asctime\": \"Fri Feb 28 09:47:54 2025 RTZ\"\r\n },\r\n \"device\": {\r\n \"name\": \"/dev/sda\",\r\n \"info_name\":
\"/dev/sda\",\r\n \"type\": \"scsi\",\r\n \"protocol\": \"SCSI\"\r\n },\r\n \"scsi_vendor\": \"Intel\",\r\n \"scsi_product\": \"Raid 1 Volume\",\r\n
\"scsi_model_name\": \"Intel Raid 1 Volume\",\r\n \"scsi_revision\": \"1.0.\",\r\n \"scsi_version\": \"SPC-3\",\r\n \"user_capacity\": {\r\n \"blocks\": 111357952,
\r\n \"bytes\": 57015271424\r\n },\r\n \"logical_block_size\": 512,\r\n \"scsi_lb_provisioning\": {\r\n \"name\": \"thin provisioned\",\r\n \"value\": 2,\r\n
\"management_enabled\": {\r\n \"name\": \"LBPME\",\r\n \"value\": -1\r\n },\r\n \"read_zeros\": {\r\n \"name\": \"LBPRZ\",\r\n \"value\": 0\r\n }\r\n },\r\n
\"rotation_rate\": 0,\r\n \"logical_unit_id\": \"0x61fa116d01000000001517ffff0aeb84\",\r\n \"device_type\": {\r\n \"scsi_terminology\":
\"Peripheral Device Type [PDT]\",\r\n \"scsi_value\": 0,\r\n \"name\": \"disk\"\r\n },\r\n \"smart_support\": {\r\n \"available\": false\r\n },\r\n
\"temperature\": {\r\n \"current\": 0,\r\n \"drive_trip\": 0\r\n },\r\n \"seagate_farm_log\": {\r\n \"supported\": false\r\n }\r\n}\r": exit status 4.

The data is coming, but it does not have the correct format.

Ubuntu 24
zabbix server 7.2.4
zabbix agent2 7.2.3
smartmontools 7.4
template is SMART by Zabbix agent 2

Help me please

[evert]

The company cannot afford to reduce security and publish such instructions, but we can)

Does the method you describe reduce security? In what way?

[PavelZ]

The company cannot afford to reduce security and publish such instructions, but we can)

Old template also have a good example in readme :

Cmnd_Alias SMARTCTL = /usr/sbin/smartctl
zabbix ALL= (ALL) NOPASSWD: SMARTCTL
Defaults!SMARTCTL !logfile, !syslog, !pam_session 

The commands will continue to run, they will just clutter the logs.