WD Caviar Green 3TB, SMART, hdparm and LCC

Hi,

My NAS setup is made of :

• a SuperMicro X7SPA-HF-D525 motherboard,
• 4 GB of RAM,
• 2 WD30EZRX 3 TB 3.5" hard drives for my data, assembled in a RAID 1 array,
• 1ST3750840AS (Seagate Barracuda 7200.10) 750 GB 3.5" hard drive for old backup data,
• 1 WD2500BPVT (Western Digital Scorpio Blue) 250 GB 2.5" hard drive for the OS: OMV 0.5.8.

I configured the 3 3.5" hard drives to enable APM (level 1 - minimal power draw) and to enter standby after 5 minutes of inactivity. It may seem too low, but I wanted to size the minimal power draw of the box.
I used my UPS that has an LCD display showing the current power consumption as an inacurate energy meter. The whole setup was drawing 25 W as per the UPS display.

A few days ago, I enabled SMART monitoring on my two WD30EZRX 3TB drives.
I started to see fast power draw variation. Jumping from 25 W to 38 W, slowly decreasing to 35 W, then falling to 30 W, then jumping again to 38 W. It seems cyclic, about every 20 seconds.

I thought I would just disable SMART monitoring and the power draw would be stable again.
But it didn't work. And I did press the green/rotating arrow button.
I checked the /etc/smartd.conf file. The settings had been applied.
I wondered if a reboot would help, but it didn't.
I did a 'smartctl -a' on the drives and noticed that the auto offline data collection was still enabled.
I thought that disabling SMART monitoring through the OMV web GUI would really disable SMART monitoring but it doesn't. As far as I understood SMART settings, these are applied and are kept beyond a power cycle. Thus, only removing lines in /etc/smartd.conf seems to be not enough. I may have misunderstood the whole thing, but the behaviour I've noticed seems validate my theory.
Issuing a 'smartctl --offlineauto=off /dev/sdX' on the two drives solved the issue (not really an issue) of power consumption bumps.

While fiddling with smartctl, I came accross something else: a sky rocketting Load Cycle Count. A quick search on the web provided me with some answers: WD "Green" drives park their head rather aggressively (see http://koitsu.wordpress.com/20…-aggressive-head-parking/).
As the Load Cycle Count stopped increasing as soon as I disabled the SMART monitoring _and_ the auto offline data collection, I re-enabled SMART monitoring so that I could investigate on the LCC issue.
I tried to change the APM settings of the drives, hoping for a change, but whatever the value I chose, even Disabled, nothing changed: the LCC still increased. As for the SMART monitoring, I thought a reboot could solve the issue, but it did not...
I looked at the /var/log/boot file and noticed the following lines :

Setting parameters of disc:  /dev/sda.
        /dev/sdb failed!
        /dev/sdc failed!
        /dev/sdd.

Disks sdb and sdc are my 2 WD Green hard drives.
So I tried several values for the APM level for these two disks using the CLI: hdparm -B<value> /dev/sdX. None of the values I tried worked.
I, then, checked the hard drive's capabilities using smartctl and saw that these drives do not support the APM, nor the AAM.
By the way, the webgui, when applying the parameters changes did not yield any warning or error, as far as I searched (even through the log files). I think the webgui or the logs should have mentionned the error. Event better, the APM and AAM fields of the Physical Drive Edit form should only be enabled for drives that advertise these features. Shouldn't it ?

So, here am I, with two drives I cannot change the APM settings, parking their heads way too often. As stated by the author of the link referenced above, these drives are using an internal timeout of about 8 seconds to trigger the heads parking. With SMART auto offline collection starting about every 20 seconds, it is about 3 Load Cyle per minute, thus about 4300 Load Cycle per day, more than 1.5 million per year... Way too much !

I just wanted to share my experience and ask for comments. All in all, I will now stay away from these WD Green drives if I want SMART monitoring. I think I'll replace them (any advice will be welcomed !).

Pierre

PhantomSens: I'm looking into these RED drives right now. When I first heard about them, I was fearing a marketing effect. I will take your advice.

davidh2k: As stated, it was only set this way so that I could size the minimal power draw of the whole box. Right now, the spin down timeout is disabled. But thanks for the advice

Based on what you two said, I think I'll have to go buy some Reds and use the Greens for backups only. That way, they'll be in standby most of the time, only spinning up once a day (as far as I don't enable SMART monitoring on them...).
Thank you !

Pierre

From what I understood, it is not SMART reading that wakes the disk up. It automatic offline data collection. And I'm sure it is that, that "loads" the heads.
I've asked WD support about that and here is their answer :

• automatic offline data collection starts after 8 seconds of inactivity;
• heads parking takes place after 20 seconds of inactivity.

So, the time-line of events seems to be the following :

• 0 sec - heads are parked;
• 8 secs - auto offline data collection starts and heads are loaded;
• 20 secs - as heads where loaded and as auto offline data collection is not considered as a regular disk activity, thus the last 20 seconds are considered as inactivity, so they're parked again, and auto offline data collection is interrupted.

Support says I shall not fear about the LCC crazy increasing, the warranty of the Green drives is of 2 years, and the 300,000 LCC limit advertised for these drives is not realistic, the heads' arm's mechanism is tailored to go well beyond that limit.
The conclusion of the support is that, using these drives for more than 8 hours a day is not the use they were made for, or I shall disable auto offline data collection.
By the way OMV should provide an option to disable that feature, enabling users to have SMART on their drives but not auto offline data collection. Just my 2 cents.

Pierre

Lets hope so, as I'm going to buy some based on your advice

Pierre

I'll contact the WD support team then. They were really cool on the phone this morning. Moreover, their french phone number is free

Hi all,

First, I want to reply to sektion31 about the physical drive parameters not being set in the webgui. I've already submitted a bug report about that. See issues 839, 840 and 841 at http://bugtracker.openmediavault.org/.

Now, regarding -B and -S options of hdparm...
These options are completely different things.

-B is only usable on drives that have the advanced power management feature. You can check that by listing your drive capacities using hdparm -I <device>. In the "Commands/features:" section, you'll see a line for each feature implemented by your drive. A star before the feature means that the feature is enabled. for instance, here's the output for my WD scorpio blue, on which I set the APM level to 254:

Commands/features:
	Enabled	Supported:
	   *	SMART feature set
	    	Security Mode feature set
	   *	Power Management feature set
	   *	Write cache
	   *	Look-ahead
	   *	Host Protected Area feature set
	   *	WRITE_BUFFER command
	   *	READ_BUFFER command
	   *	NOP cmd
	   *	DOWNLOAD_MICROCODE
	   *	Advanced Power Management feature set
	    	SET_MAX security extension
	   *	48-bit Address feature set
	   *	Device Configuration Overlay feature set
	   *	Mandatory FLUSH_CACHE
	   *	FLUSH_CACHE_EXT
	   *	SMART error logging
	   *	SMART self-test
	   *	General Purpose Logging feature set
	   *	64-bit World wide name
	   *	IDLE_IMMEDIATE with UNLOAD
	   *	{READ,WRITE}_DMA_EXT_GPL commands
	   *	Segmented DOWNLOAD_MICROCODE
	   *	Gen1 signaling speed (1.5Gb/s)
	   *	Gen2 signaling speed (3.0Gb/s)
	   *	Native Command Queueing (NCQ)
	   *	Host-initiated interface power management
	   *	Phy event counters
	   *	Idle-Unload when NCQ is active
	   *	NCQ priority information
	   *	DMA Setup Auto-Activate optimization
	    	Device-initiated interface power management
	   *	Software settings preservation
	   *	SMART Command Transport (SCT) feature set
	   *	SCT LBA Segment Access (AC2)
	   *	SCT Features Control (AC4)
	   *	SCT Data Tables (AC5)
	    	unknown 206[12] (vendor specific)
	    	unknown 206[13] (vendor specific)
	    	unknown 206[14] (vendor specific)

On my WD Green, the Advanced Power Management doesn't appear in that list, meaning that whatever the value you set in the webgui, even disabled or level 255, the drive won't change its behaviour (see issue 840).

Advanced Power Management is a feature implemented in the drive firmware, it can play with head parking - like in my case/issue - it can lower the rotating speed of the drive, and so on. It can even spin down the drive.

-S is a timeout that doesn't depend on the APM. This time of inactivity that will trigger a spindown. Setting a 0 timeout disables that feature.
You may have set your drive to an APM level of 1 (the most aggressive one, with standby) and set the standby timeout to 30 minutes for instance. In this case, your drive may have already spun down - depending on its design - before the 30 minutes timeout.

I hope I've been clear enough. Do not hesitate to request further details
Good evening all !

Pierre

sektion31: so did you fill a bug report for the write cache enabling/disabling feature ? For the -B -S priority, I would say that -S has a higher priority. But that's just my feeling, I cannot do the test as my only APM featured drive is my system drive, thus,issuing a command would wake it up.

KM0201: what problems did you have with running RAID1 on these drives ?

Pierre