We have got a few P3700 NVMe Gen3 PCIe SSD 800GB cards for a few of our servers, all running CentOS 7. We would like to minitor their temperature. As far as I can tell, the current smartmontools http://sourceforge.net/p/smartmontools/code/HEAD/tree/trunk/smartmontools/drivedb.h do not support NVMe devices. So, I went ahead and installed https://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=21099 isdct on our servers. But other than reporting TempThreshold: 85 degree C, I didn't see any option from the Intel tool for the current drive temperature. Is there such a software tool available at all?
BTW, why the isdct states TempThreshold: 85 degree C, but https://www-ssl.intel.com/content/www/us/en/solid-state-drives/ssd-dc-p3700-spec.html its spec states (for the AIC), 0 - 55 deg C? Also, if I run smartctl -d scsi /dev/nvme0n1, it does report Current Drive Temperature, can I believe it?
Hi Zack, the Intel® SSD Datacenter Tool can dump the temperature to a file.
This is the syntax for the command:
Isdct dump –destination -intelssd datatype=nvmelog logid=197
You may use the filename of your choice.
The value for can be found by using the show option for isdct as explained below:
First, obtain a list of installed drives:
Isdct show –intelssd
Find the Index number of the drive you want, and then issue the Dump command. Assuming the drive index in question is 0, the command would be:
Isdct dump –destination temperature.txt -intelssd 0 datatype=nvmelog logid=197
I included a screenshot of the commands.
I have seen a couple of comparisons, not rigorous tests, where Intel tools and smartctl have reported the same temperature. I don't think smartctl is perfectly coded to use nvme devices but in this area you may have success using smartctl. I would be smart to do your own rigorous comparisons.
I just tried getting the temperature of a P3700 card using the command "Isdct dump –destination temperature.txt -intelssd 0 datatype=nvmelog logid=197" which works. I have a few questions though
1. If I want to monitor the temperature over a time period in secs, mins and hours, how do you advise I do that as the command just displays the current temperature and dumps it into a file.
2. Reading the temperature stats on a graph
The Intel® SSD Data Center Tool does not include the functionality to monitor the temperature on a schedule, and it does not have the option to present the results in a graphical view. You would need to use additional software to achieve this. Here are a few aspects regarding this topic for your consideration:
- You can use the ISDCT commands in an automated script, and run it on a scheduled task (cron job).
- ISDCT can save the command output in text, JSON or NVMXML formats. You can then integrate the output into other files at your convenience.
For more information about this topic, you can check the https://downloadmirror.intel.com/23931/eng/Intel_SSD_Data_Center_Tool_2_2_x_User_Guide_331961-004.pd... Intel® SSD Data Center Tool 2.2.x User Guide.