WD Green LCC count - some numbers

Status
Not open for further replies.

Pitfrr

Wizard
Joined
Feb 10, 2014
Messages
1,523
Hi,

I'm using the following system:
  • Motherboard: Supermicro X9SCM-f-o
  • CPU: Intel Xeon 1220Lv2
  • RAM: 2x8Go Kingston ECC KVR16E11/8I
  • 6x2TB HDD (RAIDZ2)
  • FreeNAS version 9.2.1.8
  • Using CIFS
I have 3 WD green (WD20EZRX) in this system and I was reading Cyberjock's thread "Hacking WD Greens(and Reds) with WDIDLE3.exe" with great attention.
I was concerned about the Load Cycle Count (LCC), so I decided to use WDIDEL3 on those 3 disks.
But I was curious about the effect it would have on the numbers coming out therefore I made some tests and thought I could share the results on the forum.


So I have 6 disks: 3 seagate and 3 WD green.
In the results I'll only take one seagate as reference and the 3 WD green.


The first numbers I started with were:
- for the seagate:
Start stop count (SMART id 04): 151
Power on hours (SMART id 09): 13600 hours
Power cycle count (SMART id 12): 61
LCC (SMART id 193): 367
Temperature: 44°C​

- for the WD Green:
Start stop count (SMART id 04): 2
Power on hours (SMART id 09): 29 hours
Power cycle count (SMART id 12): 2
LCC (SMART id 193): 766
Temperature: 41°C​

The first thing I did was to cool down the seagate drives! ;-)
(here only one of the seagate is shown but they were around 44-46°C)
I'll have to do something for the WD as well since they are quite close to 40-42°C.

Rather than giving a table full of numbers, I tried to put that in a graph.
The X axis shows the hours between each samples and the Y axis shows the LCC hourly rate.

LCC rate.png



I had the system running for some hours and got the following results:
- for the seagate, an LCC increase of 0 LCC/hr
- for the WD, an LCC increase of 25/35 LCC/hr (regardless from the APM settings)

First I tried to play around with the Advanced Power Management settings (in the FreenNAS GUI) of each drive (for the WDs).
I set two WD drives with APM at 254, one WD drive with APM at 192 and didn't notice any change in the LCC rate (within 48 hours).

Then I changed the LLC settings on the WD drives (using WDIDLE3).
I kept one drive unchanged (as reference, i.e. at 8s), one disabled and one set at 300s.

I immediately saw the difference:
- the two drives with the LCC timer changed dropped to 0LLC/hr
- the reference WD drive (with the LLC timer at 8s) continued with a rate of 35LCC/hr


Some thoughts:
- I'm surprised that both drives (with LLC timer disabled and LCC timer set at 300s) droped at 0LCC/hr. I would have expect eventually some differences in the numbers...
- I don't explain the drop at around 15LCC/hr (for all the WD drives) when the LCC timer has been changed (see on the graph).
- I don't explain the increase in the LCC rate for the drive with the timer set at 8s (the two last samples show an LCC rate over 55LCC/hr in the last 3 days).


As a conclusion: I'll definitely change the last WD disk's LCC timer! ;-D
 

Yatti420

Wizard
Joined
Aug 12, 2012
Messages
1,437
It hurts performance and the drives for 8s cycles.. They basically constants load/ unload without wdidle.. I disabled mine.. I also have mixed pool atm

Sent from my SGH-I257M using Tapatalk 2
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,176
Assuming the system dataset (almost typed "systemd ataset", which would've instantly started a systemd flame war in a Linux forum) is on the pool, the constant activity on it is most likely enough to ensure that the drives never actually park their heads, even with the timer set to 300s, explaining the numbers.
 

cyberjock

Inactive Account
Joined
Mar 25, 2012
Messages
19,526
I appreciate your time and effort in putting the info together, but your numbers are only representative for your setup. Every time you hit the timeout and park the head the LCC increases. Likewise if you have some kind of load that accesses the disk even a millisecond sooner than the LCC time, then the LCC will never increase.

So something as simple as streaming a very high bitrate movie can result in no LCC increase. Likewise a pool that is completely (100%) quiet for 24+ hours will also not see a LCC increase.

But, you did validate that the problem most people have applies to you. For the most part a typical server load is enough to create a lot of LCC increases because of how the workload works out.
 
Status
Not open for further replies.
Top