DB-유

Expanding Synology Storage by Adding Drives

· synology, storage, raid

I buy NAS units with half the bays empty on purpose, and "later" always arrives sooner than planned. There are two ways to get more space: fill the empty bays, or swap existing drives one by one for bigger ones. Both work online on DSM 6.2 without moving data off, but the RAID type decides what's allowed and how long it takes. Here's how I do both, and what I check before pulling anything.

Have a current backup first. Expansion is normally safe, but a second drive dying during a multi-day rebuild is not a theoretical risk. Storage Manager → Overview should say Healthy with no SMART warnings, and buy CMR drives if you can; SMR drives make rebuilds miserable.

What your RAID type allows

SHR / SHR-2     add drives of any size; replace with larger drives; grows with the smallest pair rule
RAID 5 / 6      add drives of equal or larger size; replace all drives to use larger capacity
RAID 1          replace both drives with larger ones; can migrate to RAID 5 by adding a third
RAID 0 / JBOD   cannot be expanded by adding drives (JBOD can, but without any protection)
Basic           single disk; can migrate to RAID 1 by adding a drive

Check yours in Storage Manager → Storage Pool. SHR is the most flexible, which is why I recommend it on units with mixed drive sizes.

Adding a drive to an empty bay

Slide the new drive into an empty tray while the NAS is running (all current models are hot-swappable). Storage Manager → HDD/SDD should list it as Not initialized within a few seconds. Then:

Storage Manager → Storage Pool → select the pool → Action → Add Drive → tick the new drive → Apply.

DSM starts a parity consistency build. The pool stays online and fully usable, but slower. Check the state from the shell:

cat /proc/mdstat

You will see something like md2 : active raid5 sda5[0] sdb5[1] sdc5[2] sdd5[3] with a reshape = 12.3% line and an estimated finish time. Reshaping 4 TB drives takes roughly 12 to 24 hours; do not power off during it.

Expanding the volume

On DSM 6.2 with a pool of type "Multiple volumes on RAID", the pool grows but the volume does not; you must expand it separately: Storage Manager → Volume → select → Action → Configure → set the new size or click Max. With "Single volume on RAID" (the default for SHR pools created by the quick wizard), the volume grows automatically after the reshape finishes.

Confirm with:

df -h /volume1
sudo btrfs filesystem show /volume1

Replacing drives with larger ones

With no free bays, replacing drives one at a time is how you grow on SHR. The sequence for each drive:

  1. Storage Manager → HDD/SSD → select the drive to replace → Action → Deactivate (this cleanly removes it; the pool becomes Degraded).
  2. Pull it, insert the larger drive.
  3. Storage Manager → Storage Pool → Action → Repair → choose the new drive.
  4. Wait for the rebuild to reach 100 percent. Only then move to the next drive. I'm strict about this; the temptation to start the next swap "while it's rebuilding anyway" is exactly how arrays get lost.

With SHR, capacity increases as soon as two drives of the larger size are in place (the parity needs a partner). With RAID 5 you must replace every drive before any extra space appears, and then expand the volume as described above.

Rebuild speed can be raised at the cost of client performance:

echo 200000 | sudo tee /proc/sys/dev/raid/speed_limit_min

This resets at reboot, which is fine.

Changing the RAID type (optional)

Going from Basic to RAID 1, RAID 1 to RAID 5, or RAID 5 to RAID 6 happens through Storage Manager → Storage Pool → Action → Change RAID Type, after adding the drive needed. SHR to SHR-2 works the same way. You cannot go from RAID 5 to SHR; that needs a rebuild from backup.

Is it really done?

Storage Manager → Overview must return to Healthy, and Storage Pool must show the new total capacity with all drives listed as Normal. From the shell:

cat /proc/mdstat
sudo mdadm --detail /dev/md2 | grep -E 'State|Active Devices|Array Size'

State : clean and the expected number of active devices means the array is complete. Run an Extended SMART test on the new drive from HDD/SSD → Health Info → S.M.A.R.T. Test before considering the job done, and schedule Storage Manager → Storage Pool → Data Scrubbing to run monthly so latent errors are caught.

Odd cases

  • The new drive appears with "Unsupported drive" or a warning about compatibility: DSM checks the model against its list but still allows use on most models; on some enterprise units it blocks it. Update DSM first, as the list ships with updates.
  • Add Drive is greyed out: the pool is RAID 0 or RAID 1 (RAID 1 wants Change RAID Type instead), or a reshape is already in progress.
  • After replacing two drives with larger ones the SHR pool did not grow: the two new drives are of different sizes, or the volume needs manual expansion. Check Storage Manager → Storage Pool for "Unallocated" space.
  • mdstat shows recovery stuck at the same percentage for hours and the drive LED blinks orange: the new drive has bad sectors and the rebuild is retrying. Check dmesg | grep -i 'I/O error'; replace the drive.

Patience is the whole trick. Let each rebuild finish.