A studio in Nansha sent a message from Synology DS920 +(4 disk positions, loaded with 2 pieces of Western Digital Red disk 4TB group SHR): storage space is only 40GB, Lightroom gallery has been unable to write in. They want to change two pieces of 4TB to two pieces of 8TB, asking us if we can handle it, and "when changing the disk, the data will not be lost".
Conclusion first say:SHR supports disk by disk hot plug expansion, data will not be lost, but the process lasts for dozens of hours, any step of chaos may roll over.We record throughout as follows.
Before you do it: three iron rules
- Backup first and then expand.During the array rebuild, the hard drive was constantly loaded and the old disk could die at any time. We rsynchronized the most critical 1.2TB slice and customer original slice to a removable hard drive, which took 9 hours. Rebuild array without backup = bet all the data on the life of two old disks.
- The new plate first physical examination.Two pieces of Western Digital Red Disk Plus 8TB (WD80EFZZ) get to the hard drive box first, full slow scan (Victoria full read test), confirm that there are no bad sectors and then PC build. The new disk is sick, and it will be more troublesome to rebuild half of the error.
- Power-off protection.There was no UPS in the studio, so we borrowed a Taishan special TG-BOX 850VA to connect NAS and router. This decision proved to be life-saving-the next afternoon really tripped once.
Expansion process: only move one disk at a time
SHR expansion rules areReplace one disc after another, and replace one disc only after the array is completely repaired.:
| step | operation | time-consuming |
|---|---|---|
| 1 | Hot-unplug drive 1 (4 TB), insert new 8 TB drive, Storage Manager → Hard Drive → Operation → Replace Hard Drive | 2 minutes |
| 2 | array enters the "degraded" state and automatically starts repair (rebuild data to new disk and expand partition) | 22 hours |
| 3 | After the repair is complete and the status returns to "healthy", repeat steps 1-2 for disk 2 | 24 hours |
| 4 | After both blocks are replaced, Storage Manager → Storage Pool → Operations → Extend File System (if not automatically triggered) | 2 hours |
The total time is about 48 hours. NAS can read and write normally during this period, but the performance is obviously reduced (IO is occupied by reconstruction). The photo studio's retouching export is twice as slow in these two days, which is expected.
Accidental: Tripping midway through reconstruction
When the second disk was rebuilt to 61%, the studio air conditioner tripped, NAS supported UPS for 12 minutes and shut down normally (we equipped "enter safe mode when power is off" in DSM). After power on, DSM automatically continued to rebuild, from 61%.No missing data, no false reports.. This is the breakpoint continuation ability of SHR/RAID reconstruction--but only if it is normally shut down. If it is a direct power failure (no UPS), there is a certain probability that array metadata is inconsistent, then you need to manually check and repair it with mdadm, the risk is much greater.
Capacity expansion completed: Why is the capacity 7.2TB instead of 14TB?
The customer was stunned to see that the total capacity after expansion was only 7.2TB. Explain:SHR Two disks of the same capacity = RAID1 mirror image, one disk capacity is used to store data, the other is redundant. Two 8TB disks can actually be used about 7.2TB (nominal 8TB = 7.28TiB, and then remove the system partition). To double the capacity, either add a third disk (SHR three disks start to provide "capacity maximization"), or change to SHR-2/RAID5 and expand disk by disk. Redundancy is not a waste, but a pre-discount for data recovery costs-4TB cleanroom platter recovery costs RMB 3000–8000, and the redundancy cost of mirror disks is much lower than this.
How to read SMART data: Can the old disk replaced still be used?
Two 4TB red disks were retired before we read SMART:
- Disk 1:26000 hours of power (about 3 years uninterrupted), number of reallocated sectors 0, sectors to be mapped 0--healthy, turned to cold backup disk.
- Disk 2: Number of reallocated sectors 12, sectors to be mapped 3--Physical Bad SectorsThis disk can only be scrapped. That is to say, if this time it is not active expansion, disk 2 may not last this winter. At that time, SHR will be degraded, and disk 1 will have problems again, which will be a data disaster.
This is why we recommend NAS users to check SMART once every six months: 05 (reallocated sector), C5 (to be mapped), C6 (irreparable) as long as the three values are not 0, it is time to prepare for disk replacement.
Attachment: SHR expansion time comparison table (DS920 + measured reference)
| Original capacity → new capacity | Single disk reconstruction time | Total time (2 sets) |
|---|---|---|
| 4TB → 8TB | 22-24 hours | about 48 hours |
| 4TB → 16TB | 40-46 hours | about 4 days |
| 8TB → 16TB | 38-44 hours | Approximately 3.5 days |
Rebuild time is proportional to disk capacity and inversely proportional to NAS performance. During this period, please keep power stable, do not restart NAS, and do not run a large number of read and write tasks at the same time.
ACCPC provides a full range of NAS services: disk expansion and replacement agency (including data verification report), array repair, misdeletion recovery, ransomware decryption evaluation, NAS system migration. Nansha Panyu on-site, other areas can pay for repair. Expansion agency service fee: RMB 600 (including 48 hours of monitoring) within 4 disk positions, data backup is calculated separately.