The hard drive cannot be recognized when the NAS is turned on. In the end, it was found that it was not the fault of the drive at all.

Last month, I cleaned out the dust on my N100 small console at home. I took it apart and blew out the dust, put it back together and turned it on. A red letter appeared on Feiniu's storage space page.: Storage pool is downgraded. My 4T CoolWolf, which is the main drive that holds more than 40,000 photos, disappeared directly from the hard drive list.

My heart skipped a beat at that time, and my first reaction was that the disk was hung up. I almost placed an order for a new disk on the spot. Later I resisted and spent half the night checking them in order. In the end, the answer was: The SATA cable is loose. The plate itself is very good. After half a day of hard work, I have completely sorted out the troubleshooting order for the problem of "NAS not recognizing the disk". There are four root causes, each with its own symptoms. Just follow the correct ones.

First distinguish the three symptoms, and then touch them with your hands

Not being able to recognize the disk is not one kind of disease, but three kinds of diseases. Don't mess with it until you get started, and make sure you know which one you belong to.

By the way, if you can't connect to the entire NAS and can't open web pages, that's another matter. It's a problem with the machine itself or the network. I wrote an article before Troubleshooting NAS suddenly inaccessible , save the machine first and then talk about the hard drive.

My situation is a mixture of the first plus the third: This disk is not in the list, but because it is in the storage pool, the pool directly reports it for downgrade. This is the scariest, but also the most common.

After categorizing symptoms, don’t rush to type commands. Put your fingers on the hard drive first, turn on the computer, listen and touch. You can clearly feel the moment when the 3.5-inch mechanical disk is powered on and starts spinning. A fine vibration comes from the disk body, and there is also a very light motor sound.

This splits the problem in half:

The disk is spinning , indicating that the 12V power supply has arrived and the motor is working. The problem is likely to be on the data link - SATA data cable, interface, disk slot. Check in these directions.

The disk does not spin , there was no vibration at all, that’s because the electricity didn’t arrive at all. Check whether the power cord is plugged in tightly, try changing the power connector, and then suspect the power supply itself. When I was cleaning the dust, I unplugged the power cord. When I plugged it back in, I didn't press the big 4-pin all the way. It looked like it was plugged in, but it was actually a virtual connection.

There is a third kind of touch that you should be wary of: The disk is spinning, but there is a regular "click-click" sound, and the seek sound is ridiculously loud. This is probably a matter of the magnetic head or the disk. Stop messing with the cables and think about what to do with the data quickly. This will be discussed later.

SATA cable is the number one habitual offender, it was the one that caused my car to crash during cleaning.

The disk is spinning but not in the list. 90% of it is the data cable. SATA cables are so cheap that no one pays attention to them, but they are the most vulnerable link in the entire link.

My N100 uses loose wires given by my boss during installation. There are no metal buckles, just bare plugs. Normally it's fine if I don't move it. I cleaned it, pulled it out once and then plugged it back in. I thought it was tight, but it was actually half a millimeter off. After turning on the system, the system can occasionally scan, sometimes cannot, and finally disappears completely - this is a typical bad contact. When I looked at the logs in SSH, I saw a full screen of link down, the link was repeatedly disconnected and reconnected, and the driver gave up in the end.

Remember three points when changing lines:

First, buy a wire with a metal buckle. When inserted into place, there will be a soft "click" sound. After the buckle is locked, it will not loosen even if it is transported. Bulk unbuckled wires cost two or three yuan each, and buckle wires only cost five or six yuan, so don’t save money.

Second, don’t buy ones that are too long. In the SATA specification, it is recommended that the data cable be within one meter. If the cable is too long, the signal integrity is not guaranteed. A 50cm cable in the chassis is more than enough.

Third, take a look at the interface before plugging it in. The SATA plug is an L-shaped fool-proof port. It cannot be inserted in the wrong direction, but the plastic interface can be cracked if it is tilted and pushed hard. If the gold finger is oxidized and blackened, rub it gently with an eraser twice before inserting it.

After I replaced a cable with a buckle, the Cool Wolf came back immediately. The storage pool automatically restored itself after ten minutes, and there was no shortage of more than 40,000 photos.

The account of power supply: During the first one or two seconds of startup, a 3.5-inch disk consumes 25 watts.

If the line is still intermittent after changing the cable, or if the machine with multiple disks has disks disappearing when it is cold-started, then the power supply needs to be considered.

A 3.5-inch mechanical disk only consumes 5 to 8 watts when running normally. When I measured the power consumption of the whole machine before, I made a mistake. When reading and writing, the extra power consumption of each disk is only a few watts, which is not noticeable. But the moment it starts up is something else entirely.: When the motor is pulled from standstill to 7200 rpm, the 12V current will rush to nearly 2 amps, which translates into 20 to 25 watts per disk, and lasts for one to two seconds before falling back. The 12V 6A adapter that comes with my N100 has a total of 72 watts. It comes with a mechanical disk and a system disk, so the margin is enough.

My cousin's second-hand Synology with four bays fell on this. He put four 3.5-inch disks in one after another. When the computer was cold booted, all four disks started spinning at the same time. The optical hard disk consumed nearly 100 watts in an instant. Adding the motherboard and the power supply were struggling for a second or two. The performance of being unable to bear it is particularly typical.: Occasionally a disk cannot be recognized, just restart it and it will be fine. It is more likely to happen when the weather is cold. He himself was wondering if the disk was broken, so he took it to the computer to test and found that the disk was fine.

There is a very economical way to solve this critical power supply situation.: Before starting the machine, unplug one disk, wait until the machine is fully started and the other disks have stabilized, and then hot-insert the disk back. Finished machines such as Synology support hot-swappable devices, and the SATA protocol itself has hot-swappable capabilities. The price is that you have to do it manually every time you cold boot, which treats the symptoms but not the root cause, but it can save the emergency.

The root cause is either to replace the high-power power supply (the finished NAS depends on the official accessories), or to control the number of disks. Calculate this account before buying. Don’t imitate me. How to choose NAS hard drive When I was watching the disk, I forgot about the feeling of the power supply at home.

Trays and backplanes: Cross-replacement, you will know whose fault it is as soon as you change it

Even after changing the wires and power supply, there is still no problem, but the disk is still not recognized. At this time, we need to use the cross method to eliminate the suspects one by one. The logic is very simple: Swap the disk and disk positions to see which fault will follow.

Insert the missing disk into another disk slot that is confirmed to be normal. If it is recognized, it means that there is no problem with the disk. It is a matter of the original disk slot, the cable or the interface. Insert a normal disk into the faulty disk. If it also disappears, it is definitely a problem with the disk.

In a finished machine with multiple bays, there is a backplane behind the bays. The bays are not plugged directly into the motherboard, but pressed against the contacts on the backplane. For machines that have been used for several years, oxidation of the backplane contacts and fatigue of the spring blades may occur, which may mean that a certain disk slot is particularly prone to bugs. All I can do is re-plug and unplug a few times and rub the contacts. If that doesn't work, it's a hardware failure on the backplane or motherboard interface, which is beyond the scope of home repair.

There is an old pitfall in setting up a self-organized NAS. By the way, I would like to mention that: If the SATA mode in the BIOS is changed to IDE or RAID, various strange phenomena will occur when the system recognizes the disk, confirming that it is in AHCI mode. Black Qunhui players should be familiar with this article.

Not over yet: Two lines of commands in SSH to see the truth, when is it really the fault of the disk?

The above are all physical layer troubleshooting. If your machine is still on and SSH can enter, there are two lines of commands that can directly see the original records of the system's interaction with the hard disk, which is much more efficient than guessing.

dmesg | grep -iE "sata|ata[0-9]" This line will filter out all SATA-related logs in the kernel. Focus on a few words: link down The link is down. hard resetting link The driver is retrying the link. COMRESET failed Even the retry failed. The more these words appear and the denser they are, the more it indicates that it is a problem of lines and contact and has little to do with the disk itself. If you see that SStatus is 0, it means that the physical link has not been established at all, and the disk is not connected to the bus at all. Check the line and power first.

lsblk Lists all the block devices recognized by the system. It is clear at a glance whether the disk is present and whether the device name is sda or sdb. Run once before and after the physical inspection, and compare the results to know whether the action has taken effect.

Then when will it be Pan's turn to sentence him to death? The physical layer has been completely replaced - the wires have been changed, the bits have been changed, the power supply has been confirmed to be OK, and the computer still doesn't recognize it or even recognizes it but the SMART is all red - that's when the drive is really failing. At this time, don't format it and don't mess around. There are actually signs before the hard drive breaks down , look back and see that there was a signal already ； I wrote an article on how to read SMART output Tutorial on line-by-line interpretation , just look at the RAW value. When it comes to replacing the disk, how can you move the data on the old disk? Complete migration process You can copy it.

My ending this time is good: A six-dollar SATA cable solved all problems. So next time the boot disk disappears, put your wallet away first and go through the following sequence.

ending: Check in this order, don’t skip steps

One sentence summary: Among those who fail to recognize the failure of the disk, the disk itself is in the minority, and the wires and electricity are the major contributors. Let’s replace the six-dollar cable first, and then talk about anything else.