Nowadays, NAS servers have the possibility of installing hard drives for storing information, and also SSD drives (either in 2.5 »format with SATA3 interface and even in M.2 format with SATA3 or NVMe interface). Knowing the state of the discs at all times is essential, and not just their current temperature. Running SMART tests on a regular basis is critical to catching potential disk errors before it’s too late. Today in this article we will talk about how to perform SMART tests on a server, and also when we should do them.
Hard drives and SSD drives are an essential part of a NAS server . In the HDD we will store the information, normally redundant making use of different RAID levels such as RAID 1, RAID 5 which is one of the most used, and also RAID 6 and even RAID 10. In addition, we will also use the SSD as SSD cache to accelerate enormously random data reading and writing, both from the network to the NAS (write cache), and from the NAS to the connected equipment (read cache), of course these SSDs will also be configured with different RAID levels to protect the integrity of the information in the event of a physical failure of one or more SSDs. However, storage pools could also be configured with SSD disks only, with the aim of storing here information that we access almost continuously. For example, if you make use of virtual machines on your NAS server, it is a very good option to create an SSD storage pool for it, since it will accelerate the load of the operating system enormously.
With the passage of time, both mechanical hard drives and SSDs are prone to failure, if we are not attentive, we can get scared and lose one or more of the drives in a short period of time. For this reason, there are SMART tests that will allow us to obtain very valuable information about the condition of the disks, and we can replace them before they can fail.
This is not exclusive to NAS servers, a PC or laptop can also take advantage of it. Any equipment that incorporates a HDD or an SSD, we can perform a SMART test to check the health of the disk, if there is a damaged sector in which we cannot write, or if there is any other type of problem, in order to anticipate a hardware failure of the disk itself.
What is SMART and why is it so important?
SMART technology stands for Self Monitoring Analysis and Reporting Technology or translated ” Self-Monitoring Analysis and Reporting Technology .” The purpose of SMART technology is to check hard drives and SSDs to detect possible failures in them in advance, in order to anticipate a possible hardware failure, and replace the disk before having an irrecoverable loss of data.
An early detection of disk failures can allow the user to make a backup copy of its content, or replace the disk with another if we have a NAS server with a RAID level that allows this without losing data. This way we avoid an irrecoverable data loss. We must remember that a RAID is not a backup, backup copies must be made following the 3-2-1 system, make 3 backup copies, on 2 different media, and that 1 copy is offline.
If we want to use this SMART technology, two requirements have to be met:
- That the BIOS / UEFI of the computer is activated that option.
- That the operating system supports this functionality (currently all OS support it).
- The hard drive or SSD itself is compatible with it.
Currently, this option is enabled by default on computers, and hard drives as a general rule also support it. Among the most important parameters that it controls we have:
- Disk temperature.
- Data reading speed.
- Reallocated sector counter: they are the ones that have failed, and depending on the number, it can be a symptom that the unit can crash at any time.
- Use of error correction code (ECC) and error counting.
As for the values of the SMART attributes they are from the number 1 to 253, in which 1 would be the worst value. In this sense, normal values are between 100 and 200. Therefore, by performing SMART tests on our hard drives and SSDs, we can obtain a lot of information about their health.
How to view the health status of a disk in Windows
Before we dive into how to see the SMART status on a QNAP NAS server, we are going to teach you how to see the health of the disks with the Windows operating system, you can also check this if you use a Windows Server. It provides us with important information and thanks to it we can take the appropriate measures:
- Do nothing if the state is good.
- Make a backup and replace the HDD if the condition is bad.
To carry out this task, we need software that shows us information about the status of our disks. In this case, we are going to start using Crystal Disk Info that you can download from here . We have opted for the standard ZIP edition that does not require installation. If we have a hard drive in good condition, it will show us this information:
That means this particular HDD is fine, and does not require any special attention. The state of health is good and is in blue. However, when these results appear, the problems begin:
When the state is put “at risk” we should start to worry, it is time to start thinking about making a backup copy of all the data and proceed to change the HDD. When it is red, the situation is usually critical, and you may not be able to avoid data loss.
How to perform a SMART test on my NAS server
In our test lab we have a QNAP TS-1277 NAS server, which has a total of 8 3.5 “disk bays, 4 2.5″ disk or SSD bays, and two M.2 SATA3 slots. To see the status of the disks, we must go to ” Storage and snapshots “, and on the left side we will go to ” Storage / Disks / VJBOD “. Once here, we will select the disc and click on « Condition of the disc «, the following will appear:
In the «Condition of the disk» we can see the hard disk or SSD model, the capacity, the temperature, the power-on time and other information. The important information is in the “SMART Information” section, where we can see each and every one of the parameters that this important function is capable of measuring. QNAP allows us to carry out a quick test or a complete test at any time, we simply have to go to the “Test” section and select one of the two options.
The most interesting thing is in the “Configuration” section, and this is where we can configure the high temperature warning, and we can also program the quick and complete SMART tests whenever we want. This is essential to carry out a SMART test proactively, to anticipate possible failures of a hard drive or SSD.
In the case of SSDs we will also have SMART information, in addition, we can see an estimate of the life time of the SSD, and of course we can see the temperature, the advanced SMART parameters, perform a quick and complete test on demand, and even schedule SMART tests to run automatically.
In the “Global Settings” section we can have other very interesting options related to SMART. The first one is that we can activate a SMART predictive migration to migrate all the data before one of the disks fails, and we can have data loss, this can be done on both hard drives and SSDs. Thanks to this functionality, the operating system will automatically anticipate a possible disk failure before it happens.
Another option is the SMART polling time, to query the disk every so often for SMART errors. An important point is that checking too frequently could degrade the performance of the disks slightly, therefore it would be advisable to do it every hour as this is the maximum that QNAP allows.
If we are using an operating system like XigmaNAS that is specifically geared towards NAS servers, we will also have a lot of SMART information to visualize and make a decision. If we click on « Diagnostics / Information / SMART » we will be able to see the current status of SMART of all the disks that we have, in addition, with each attribute it will mark a certain value, and also a description of what each of these attributes means.
At the bottom we can see the record of the different SMART tests carried out, both short (fast) and extended (complete), in the following screenshot you can see how there is a reading error, the affected logic block (LBA), and also the lifetime of the hard drive.
If we look at the status of another of the disks, we can see how this same test has been completed without errors, that the usage time is clearly lower, and that we do not have any LBA errors. That is to say, the previous disc is a little “touched” and it would be advisable to replace it as soon as possible, however, this disc without errors is perfectly fine.
Of course, the SMART of the XigmaNAS operating system can also be configured to perform these tests on demand or scheduled whenever we want.
How often should you take a SMART test
It is very important to program the operating system of the NAS server to perform quick SMART tests periodically, and also to program complete SMART tests to the different disks that we have on the server. When performing SMART tests on a NAS we must take a series of considerations:
- It is recommended not to be making backup copies at that time.
- Try to leave the server with little or no workload.
A quick SMART test can take less than 2 minutes to perform, during this time, the read and write performance of the disks where we are doing it can be affected. In the case of full SMART tests, they can take hours to perform, therefore, it is advisable to do this task at dawn where normally there will not be too much activity, although remember not to perform backups at that precise moment because the performance will be lower.
Depending on the use of the NAS server, and whether it is always on or not, it will be advisable to schedule the SMART tests further apart in time or more regularly. For example, a quick SMART test once a week could be a good start, as we will detect the main failures that may be in the disk. A full SMART test should be performed once a month, ideally in the early morning or on the weekend when no one is working with the NAS. Since we can program it to be carried out automatically, we will not have to be present while it is being carried out, but we will have to be attentive when it is finished in case there is some kind of error.