NAS reports a hard drive failure, how do I know which one it is? The tray light, serial number, and light language are all correct. Pulling out the wrong tray is an accident.

This incident happened at a friend's house last week. His four-bay Synology system pushed an alert on his phone in the middle of the night, saying that the storage pool was downgraded and the hard drive failed. The first thing he did when he got up the next day was to open the case cover and try to remove the disk. Before removing it, he sent me a photo and asked.: Is this the one that's broken? I almost lost my breath when I saw it - he was pointing to the disk in position 3, but the warning clearly said Drive 2.

Fortunately he asked. His pool was a RAID 5, and one piece was damaged but it still held up. However, if one piece was pulled out incorrectly, and both pieces were out of place at the same time, the entire pool was bricked on the spot, and all the twenty or so photos and videos had to be sent to the recovery company. Later, I accompanied him to complete the matter, from the time of alerting to the confirmation of the physical disk, which took ten minutes. I will break down what I did in these ten minutes and write it down in this article. It’s not difficult for the NAS to warn you that the disk is broken. What’s difficult is to translate “the disk is broken” into “the piece on the second seat from the left in the chassis.”

The first move: Let’s look at the lights first. The bay lights are the fingers of the NAS.

Each row of disk slots on the front panel of the finished NAS is equipped with its own indicator light. When the machine leaves the factory, it is left to the light to determine which position is in trouble. Under normal conditions, the hard disk light is an activity light. It flashes when reading and writing, and stays on or dimly lit when idle. ； The situation of the accident is different: That person's light goes out directly, stays on without flashing, or flashes wildly out of sync with other seats. My friend's car had an accident on Drive 2, which corresponds to the second one from the left. The light was off in the morning, but he didn't notice it.

The rules for lamps vary from house to house, but there is one general rule: It's the one that looks different from the others next to it. Some machines also have an independent warning light above or on the side of the tray. It does not light up normally, but turns orange to red when something goes wrong, which is more straightforward. If you are really unsure, just go to the backend and click on the action menu of that disk, look for buttons such as "Locate" and "Identify" (the names are slightly different in each version), click on it, and the light of the corresponding disk will flash specifically for you to see which disk flashes. This trick is what I think is most worth remembering - the light is speaking for the system, which is ten times more intuitive than guessing at the background.

By the way, this is a habit that should be developed in daily life as a source of information.: When positioning the NAS, leave an angle where the lights can be seen. My friend's machine was stuffed under his desk, facing the wall with the lamp, and he didn't see the lamp after the alarm went off for a whole day. The one in my house is placed next to the TV cabinet. If you pass by and glance at it, you can tell how well the four plates are doing. Buzzers and lights are the only two ways NAS can "talk without going into the background". The beeping sound in the middle of the night Here I turned over the four types of beep sounds in the dictionary, and this article is a supplement to the light.

The second move: Backstage, the serial number is used to confirm the name and seat.

The light is a quick check, but the light may fail sometimes, and it is easy to see the serial number when removing the disk tray on a multi-disk machine. Therefore, before officially starting, you must go through the serial number verification in the background. This step is the key to nailing the "disk in the system mouth" and the "seat in the chassis".

The path on Synology’s side is very short.: When you enter the HDD/SSD page of the storage manager, each disk in the list has its own disk number and serial number. Which block has an error, the status of that line will be written directly to you - what do the words "unable to access" and "unexpected removal" mean, and whether to re-insert or replace the disk, I am Storage pool downgrade article Go through them one by one. Copy down the serial number of the line where the problem occurred and compare it with the sticker on the body of the car. This will be ironclad evidence. It’s the same as Feiniu. The hard disk list in storage management also gives serial numbers, and the idea is exactly the same.

I want to explain here: Don’t wait until something goes wrong to do the serial number matching for the first time. My current habit is to copy the serial number, disk location, and model of each disk into a memo on the day of installation, and then take a photo of the disk body sticker. This time, if my friend has this watch, ten minutes will become two minutes. What's more important is to remember to update it after changing the disk - the serial number of the old disk is written on the table, and the new disk is inserted into the same slot. If something goes wrong next time, the serial number will be mismatched and it will cause confusion.

There is also a little foolproof when unplugging the disk.: Before unplugging, check the status or serial number of the light again, and check the sticker on the panel immediately after unplugging it. Confirming twice seems verbose, but it is 10,000 times better than pulling out the wrong disk and plugging it back in again - the disk in the RAID pool is repeatedly pulled out and plugged in, and the system has to re-judge the array status every time, so there is no need to add trouble to the recovery process.

The third move: Confirm the disk in SSH, the command will not lie

If you are still unsure after the first two steps, or if the machine is a self-configured NAS or the soft routing disk has no lights, go up and use SSH commands to identify it. This is the hardest trick.

lsblk -o NAME,SIZE,SERIAL,MODEL , one command lists the drive letter, capacity, serial number, and model. You already know which disk has the serial number from the background. If you check it here, you will know whether it is called sda or sdb in the system. If you want to check your health status one more time, run smartctl -a /dev/sdb , how to read the output, which columns of numbers are worth staring at, I am SMART interpretation of that article I have written a line-by-line version here, so I will only mention one sentence here.: Before making a decision, first take a look at the model and serial number at the top of the output to see if they are correct. After a quick operation on sda, the result was sdb. There is no regret for this kind of accident.

There is a pitfall in the matter of the disk talisman that needs to be revealed.: The names sda and sdb are given by the boot sequence, not by the disk slot. The drive letter may be changed by removing and inserting disks, adding disks, or changing the startup sequence. So the correct chain is the backend serial number → disk body sticker. The disk letter is just a temporary name in the middle. Use lsblk to check it before each operation. Don't trust the one you memorized last week.

What happens if you pull out the wrong disk?: There are three types of pools with completely different prices.

Why do I take "check marks" so seriously that it deserves a whole article? Because the consequences of unplugging the wrong disk are directly linked to your array type, some pools can handle it, while others overturn it on the spot. Take the three most common ones:

This table is translated into one sentence: In a single-disk fault-tolerant pool, you only have one "opportunity to remove the disk". If you remove it correctly, it will be a repair process, but if you remove it incorrectly, it will be a disaster scene. How to choose the multi-disk fault tolerance type? Comparison between SHR and RAID1 Calculated too carefully. If it really happens that the front and rear legs of two disks are completely lost, and the pool is directly "inaccessible", then don't bother with powering on and off again and again. First evaluate whether the data is valuable. If it is valuable, how deep will the water be when it is restored? Data recovery article I’ve done it before, so I won’t repeat it here.

After checking: There are three more things to do before changing the plate

Locating the specific physical disk is only half the process. Before placing an order, arrange these three items.

The first thing is to double-check whether it is really bad. Disconnection means the disk is broken, the bracket is not pushed all the way, the disk contacts are oxidized, and the SATA cable is loose, all the symptoms are "something happened to this disk". First, go through the cable and electricity set and re-plug it for a try. The process is as follows: Can't recognize the article The three-step method of touching the market. If reinsertion fails, run SMART again to confirm the red card indicator. Don’t be fooled into buying a new disk by accidental errors. The article about signs of disk failure The five signals are also worth looking back at. Many times the market will give you reminders early on.

For the second item, copy all the parameters before placing an order. When checking the serial number, write down the model and capacity. The usable capacity of the new disk must be greater than or equal to the smallest one in the pool. If you buy a smaller disk, you will not be able to select it when repairing it. ； By the way, make sure the new disk is CMR and not shingled. The article on new disk testing There is a test. It is most worry-free to buy the same model and capacity directly. There is no need to worry about "original equipment". RAID does not choose blood, only parameters.

The third thing is to copy out the important things first during the streaking period. From the time the disk is dropped to the time it is rebuilt, it can take anywhere from a few hours to a day. This period of time is handled by a single disk. For photo documents that are lost if lost, copy a copy to your computer or mobile hard drive first. While the machine is being repaired, you won’t panic if you have a copy in hand. The complete five steps to repair and rebuild, plus what you can and can't do during those nine hours of rebuilding, The downgrade article Once you have written it down in full, just follow it.

If you do the four right things on a daily basis, you will be less panicked on the day of an accident.

This time, my review friend only spent ten minutes from the alarm to the confirmation, but what really saved him were a few habits he had accumulated in his daily life. List it down for you to copy:

First, the alarm push is configured properly. He could receive notifications in the middle of the night because the push notification he had configured before was working. Otherwise, he might have been downgraded for several days before discovering it. The warning article There is a matching method. The second is to set the position to provide an angle for viewing the light, as mentioned before. Third, on the day of installation, copy the disk slots, serial numbers, and models into a table, and update it when the disk is replaced. Fourth, run SMART regularly on each disk. Whether the remapping and RAW values ​​to be mapped are zero is a matter of just a glance. Interpretation method You can learn it in five minutes.

Ending. The five words "hard drive failure" reported by the NAS are just the beginning. The real job is to turn the Drive 2 in the system's mouth into the disk that you can touch in your hand. The lamp will look at the seat and serial number to provide ironclad evidence of identity and order. After completing the three moves, you will start again. After a night of nervousness, you will become nervous again, and nothing will happen. My friend later told me that he now glances at the row of lights every time he passes by - I think this is a good habit, much better than when I would just flip through my diary after an accident.