[Help] TOS 5.0.11 (F4-212) – "Disk hibernation: disabled" puts all disks to sleep every minute
Posted: 19 Nov 2026, 06:31
Product: TerraMaster F4-212 (x3 units), TOS 5.0.11 (TOS_X642.0_5.0.11)
Summary
With disk hibernation set to DISABLED in the web UI (/etc/tos/standby.conf contains
"standby=0"), the ter_smartfan daemon sends an immediate sleep command (hdparm -Y)
to every disk roughly once per minute, as soon as the disk is idle. "Disabled" is
handled as "sleep immediately" instead of "never sleep".
Evidence (identical on all three units)
- /opt/var/log/smartfan.log repeats, about once per minute:
exec DiskSleep, since last time: 0.05 how long from idle to sleep: 0
- Observed processes, child of ter_smartfan: hdparm -Y /dev/sda … /dev/sdd
(all 4 disks within a 75-second window)
- Kernel log: hundreds of "ataN.00: waking up from sleep / hard resetting link /
retrying FLUSH" (332 in 318 h on one unit, 375 in 574 h on another).
- SMART Start_Stop_Count: 278,000–285,000 on four WD Red Plus 4 TB disks with
~9,600 power-on hours (~30 spin-downs per hour); 65,535 (counter saturated)
on the disks of a second unit.
Impact
- Very slow access: every request waits for disks to spin up.
- RAID rebuild after a disk replacement ran at ~150 KB/s instead of ~110 MB/s
(md superblock writes wait for sleeping members). Stopping ter_smartfan
immediately restored 111 MB/s.
- Heavy, avoidable mechanical wear; one disk failed during this period.
Workaround
Setting hibernation to a real value (e.g. 5 h, standby=5h) stops the behaviour
at once, without restarting anything: 0 sleep commands and 0 forced wake-ups
over 5–10 minutes, verified on all three units.
Expected
standby=0 ("disabled") must never send a sleep command to the disks.
Summary
With disk hibernation set to DISABLED in the web UI (/etc/tos/standby.conf contains
"standby=0"), the ter_smartfan daemon sends an immediate sleep command (hdparm -Y)
to every disk roughly once per minute, as soon as the disk is idle. "Disabled" is
handled as "sleep immediately" instead of "never sleep".
Evidence (identical on all three units)
- /opt/var/log/smartfan.log repeats, about once per minute:
exec DiskSleep, since last time: 0.05 how long from idle to sleep: 0
- Observed processes, child of ter_smartfan: hdparm -Y /dev/sda … /dev/sdd
(all 4 disks within a 75-second window)
- Kernel log: hundreds of "ataN.00: waking up from sleep / hard resetting link /
retrying FLUSH" (332 in 318 h on one unit, 375 in 574 h on another).
- SMART Start_Stop_Count: 278,000–285,000 on four WD Red Plus 4 TB disks with
~9,600 power-on hours (~30 spin-downs per hour); 65,535 (counter saturated)
on the disks of a second unit.
Impact
- Very slow access: every request waits for disks to spin up.
- RAID rebuild after a disk replacement ran at ~150 KB/s instead of ~110 MB/s
(md superblock writes wait for sleeping members). Stopping ter_smartfan
immediately restored 111 MB/s.
- Heavy, avoidable mechanical wear; one disk failed during this period.
Workaround
Setting hibernation to a real value (e.g. 5 h, standby=5h) stops the behaviour
at once, without restarting anything: 0 sleep commands and 0 forced wake-ups
over 5–10 minutes, verified on all three units.
Expected
standby=0 ("disabled") must never send a sleep command to the disks.