Power-off retract count / Load cycle count - Normal numbers?

[gberg]

Hi, What's  normal numbers  on HDDs?

 

I got 5 10TB WD Whites, and they look like this, do I need to worry about the numbers?

 

sde - Parity:

Power on hours: 2453 (3m, 10d, 5h)

Power-off retract count: 3109

Load cycle count: 3109

 

sdb - Disk1:

Power on hours: 2453 (3m, 10d, 5h)

Power-off retract count: 3805

Load cycle count: 3805

 

sdd - Disk2:

Power on hours: 2454 (3m, 10d, 6h)

Power-off retract count: 3669

Load cycle count: 3669

 

sdc - Disk3:

Power on hours: 2460 (3m, 10d, 12h)

Power-off retract count: 3318

Load cycle count: 3318

 

sdf - Disk4:

Power on hours: 398 (16d, 14h)

Power-off retract count: 273

Load cycle count: 273

 

 

[S80_UK]

Some drives always seem to report the same value for Power-off retract count and for Load cycle count.  On my main server, 10 out of 12 drives do this, although I never noticed until now.  Of course, the power off retract count should never be greater than the number of power cycles if it was correctly reported.

 

Your numbers are small.  It is not uncommon to have load cycle count values in tens of thousands or even hundreds of thousands in some cases.

 

I don't see anything here to be concerned about.

[gberg]

Ok, so you think this is not correctly reported, the  power cycle count is 10 for Disk 1 to 3, and for disk 4 it is 5, and as you said power off retract count should never be greater than the number of power cycles.

 

Also when I tried using  idle3-tool to check what the idle3 timer is set to I get, and I get the same error for all drives:

 

root@UNRAID:~# idle3ctl -g /dev/sde
sg16(VSC_ENABLE) failed: Input/output error

 

[JorgeB]

wdidle3 won't work on these drives, and in these models it's normal for power off retract count to be same as load cycle count and can be much higher than start stop count.

 

While your numbers are nothing to worry about, I've seen WD drives with more than a million load cycle count you might want to review your current spin down settings since the numbers do seem a little high for the power on hours, e.g. this is my WD 10TB white label:

 

Power on hours: 7564 (10m, 11d, 4h)

Power-off retract count: 398

Load cycle count: 398

 

 

 

 

[gberg]

24 minutes ago, johnnie.black said:

wdidle3 won't work on these drives, and in these models it's normal for power off retract count to be same as load cycle count and can be much higher than start stop count.

 

While your numbers are nothing to worry about, I've seen WD drives with more than a million load cycle count you might want to review your current spin down settings since the numbers do seem a little high for the power on hours, e.g. this is my WD 10TB white label:

 

Power on hours: 7564 (10m, 11d, 4h)

Power-off retract count: 398

Load cycle count: 398

Ok, all my drives use  the default spin down delay, which is set to 3 hours. What is your settings?

And yes, your numbers seems more normal to me, your power on hours are about 3 time more than mine, but Power-off retract count and Load cycle count are about 1/10 of mine.

Edited November 18, 2019 by gberg

[JorgeB]

1 hour ago, gberg said:

Ok, all my drives use  the default spin down delay, which is set to 3 hours. What is your settings?

Currently that disk is set to 1 hour, but it very rarely spins down since there are almost constant reads from it, check you syslog, if your disks are spinning down multiple times a day with a 3 hour setting it's probably a good idea to increase it.

[gberg]

4 minutes ago, johnnie.black said:

Currently that disk is set to 1 hour, but it very rarely spins down since there are almost constant reads from it, check you syslog, if your disks are spinning down multiple times a day with a 3 hour setting it's probably a good idea to increase it.

Mostly it's only the parity drive that spins down, the data drives aren't even always spinning once every day.

I've increased the timer to 5 hours and we'll see how it goes.

 

But would the spindown timer even affect this since my drives  do Power-off retract / Load cycle count about once every 30 minutes, and the numbers have increased by 10 on all drives since I started this thread about 5 hours ago?

[JorgeB]

It's not just the spin down, drives themselves will park the heads after some seconds, not sure what the value is for these, so if the drives are accessed every few seconds for a short time, and idle timer kicks in, LCC will keep increasing.

 

 

[gberg]

37 minutes ago, johnnie.black said:

It's not just the spin down, drives themselves will park the heads after some seconds, not sure what the value is for these, so if the drives are accessed every few seconds for a short time, and idle timer kicks in, LCC will keep increasing.

 

 

Exactly, so in reality the LCC might increase less if the drives would spin down more often, because as long as the drives are spun down ther will be no LCC?

[JorgeB]

Possibly, if nothing keeps waking them, worth a try.