Is there any way to stop WD green drives from going to sleep?

[Red Squirrel]

I built a server for our church to store recordings, and since performance was not needed I opted to just go with green drives. It did not occur to me that these go to sleep, which causes the raid array to crap out. Is there any way to stop this from happening? I keep getting drives dropping out, and then I realized that's probably why.

 

[hennessy1]

I believe the only way that would no result in getting the raid edition of the green drives is to constantly be using them. That is the reason they are green drives is they will go to sleep so they are not perfect for raid setups.

 

[Spikesoldier]

i hear this also happens to the WD blue and blacks, and kinda paints the picture that WD wants to strong-arm you into purchasing their premium RE products for RAID use.

 

[Blain]

I built a server for our church to store recordings, and since performance was not needed I opted to just go with green drives. It did not occur to me that these go to sleep, which causes the raid array to crap out. Is there any way to stop this from happening? I keep getting drives dropping out, and then I realized that's probably why.

What kind of a RAID array is being used?

 

[Red Squirrel]

Linux mdadm. Raid 5 with 4 drives.

 

[Blain]

Rotate Samsung F3R drives in until the whole array is rebuilt with these RAID edition models.

 

[Diogenes2]

I don't think that will keep the WD Green's from going to sleep ...:biggrin:

 

[Cr0nJ0b]

have you looked at HDPARM? I've used WD drives in an mdadm setup with no issues, but I generally like my drives to sleep when they aren't used. I think there is a way to disable sleep mode or at least put it off for a very long time. Check there.

 

[Red Squirrel]

Rotate Samsung F3R drives in until the whole array is rebuilt with these RAID edition models.

We already spent over 500 bucks on this, I'm not throwing yet more money at it.

 

[corkyg]

So, placing NEVER in Power Management has no effect on Sleep?

 

[nanaki333]

http://www.xlr8yourmac.com/tips/Disable_WDGreen_HeadParking.htm

try that. you can delay it much longer

 

[Red Squirrel]

Thanks I'll give that a try.

Also I don't think this has anything to do with power management in the OS, it's the drive itself that goes to sleep. They are made that way to save energy, though as stated in that link it does more harm than good as there is a limited amount of park operations so it ends up creating more pollution from drives failing prematurely. Best to just keep it spinning.

Really had I thought of that I would have paid a bit more and got the non green drives but too late now.

 

[Mark R]

Drives simply going to sleep shouldn't cause them to drop out of the array - at least not on linux software raid (mdadm). However, the WD caviar green drives are widely reported (but not universally) as having erratic behavior in linux due to their power saving features (presumably, because they can confuse some SATA controller chips, but I don't think the causes is at all clear)

The WD green drives power saving mechanisms are not affected by hdparm/sdparm spindown settings, which further complicates matters.

WD do provide a DOS utility called wdidle3_1_05 which can be used to alter the power saving settings. You need to transfer the drives, one at a time, to a PC capable of running DOS and run the command on each drive. see link

Some people that have reported constant dropouts in RAID have found some other workaround - e.g. disabling NCQ with hdparm. (The theory is that there is some glitchy behavior between the drive and SATA controllers that occurs when NCQ and powersaving simulateously try to do something).

Others have traced the problem not to the drives, but to faulty hardware elsewhere (e.g. faulty SATA cables).

If you do this again, do yourself a favor and get RAID certified drives.

 

[nanaki333]

friend of mine had an issue with his home raid5 array where they would spin down like every 20 seconds or whatever. after he did the "fix" i linked to, didn't happen anymore since it was set to like an hour or whatever.

 

[lehtv]

I actually stopped using my WD green, I didn't like the sound it made every time it powered up, and the delay when accessing files on it. Obviously that wasn't a problem when I was using it as my system HD.

 

[corkyg]

Any drive that is not controllable in Power Management would never find its way into any system of mine.

 

[sub.mesa]

Head parking is a good thing, since it lowers idle power drain and protects against shocks much better during the period the head is parked. The problem is that WD decided to be very aggressive with head parking, letting it happen every 7 seconds.

Instead of disabling head parking completely, you can opt to set it at a much higher value like 120 seconds. That should prevent continuous parking when the drive is regularly used, but still allows head parking during those periods where the drive is not used, but still spinned up and ready to process I/O in no-time.

Leaving the head-parking at default 7 seconds results in very high Load Cycle Counts (LCC) - which can be displayed with a SMART reader like smartmontools (smartctl -A /dev/...). For most consumer disks the rated cycle counts is between 100.000 and 300.000 but I've seen disks with a year spinning time to be as high as 400.000. This could cause the disks to fail prematurely due to excessive head parking, wearing out the mechanical components to a degree that exceeds the cycle count it was designed for.

There is no reason to use special "RAID certified" drives for a Linux software RAID array, though. Those drives, such as the Western Digital RE series, has a proprietary TLER feature which is needed with Windows FakeRAID and regular Hardware RAID, since those RAID engines typically disconnect a disk that is performing recovery. Linux and other non-Windows software RAID is not affected by this, and as such have no need for TLER/CCTL/ERC feature. Activating TLER on a Linux software RAID might even be dangerous, because it shortens the time the harddrive spends on error recovery. If your array is degraded and one disk has a weak sector, you would want to let the drive spend as much time as it needs to try to recover that sector, instead of being returned an I/O error which is what happens with TLER and its equivalents.

 

[Red Squirrel]

Any drive that is not controllable in Power Management would never find its way into any system of mine.

Yeah I should have read up more on these before buying. I had heard they were problematic in some setups but never figured it was that much of an issue.

I have a bunch of 1TB Hitachi drives I use for backups. Debating on just using those in the array instead, and using the greens for backup drives. I way oversized the array anyway. By the time the church it's for fills it, the rapture will have come.

 

[MobiusPizza]

It's a shame WD withdraw their utility of changing sleep times for Green drives

Shouldn't be hard to use say AutoHotkey to make a program which will 'ping' the drives by loading random small files from the drives to keep alive.