Western Digital Gold appears to go to sleep minutes after last use

[c627627]

I pushed this all the way to the top. I specifically asked a Level 3 Western Digital engineer with 27 years of work experience there, who called me three times, if IDLE_B mode can cause a 4 second delay.
"I don't know" was a specific answer. Where do you go from there?

There was no first hand knowledge whatsoever about this at Western Digital. It all boiled down to here's what your drive settings are. Period.

No information about WD Red was given whatsoever to me because I would need BOTH a model number *and* a serial number, to ask any questions, and if I buy a WD Red, all they would say again is "here's what your drive settings are." Period.



So, I can't go to Storage Review and the like without at least one other person who owns the same WD Gold line as me saying they noticed a 4 second delay after X minutes of last activity too.


Every single level of support tried to blame Windows.
I said I have a WD Black that is not affected, Windows cannot issue commands to individual hard drives, it's to ALL mechanical hard drives.
Level 3 tech said he didn't know and to "ask Microsoft."

Every single level of support tried to also blame Windows programs. I said, I don't have anything installed there, it's just storage, only storage on WD Gold.

I asked if my symptoms can be taken to someone else, just to get an opinion, and was told who I was talking to - the end of the WD Support Line.



So all right, the only way I would know if it's all WD Golds that do this is to find someone else who owns one, if you guys ever see anyone owning a large capacity WD Gold, ask them if they noticed a 4 second delay after a short period of inactivity. The End.

 

[ATMINSIDE]

He had me run this program

WeTransfer - Send Large Files & Share Photos Online - Up to 2GB Free

WeTransfer is the simplest way to send your files around the world

we.tl

C6, I have a small ask of you. Could you download the program and attach it directly to the forum instead of linking it? That way when the WeTransfer expires, there is still access. Thanks!

 

[Janus67]

Did you ever bother to check in another os install to see if it happens there? Completely ignoring your wd black.

 

[c627627]

C6, I have a small ask of you. Could you download the program and attach it directly to the forum instead of linking it? That way when the WeTransfer expires, there is still access. Thanks!

It appears it has already expired. The zip file is 22MB, did you want to see if you can lift the file attachment limit on my account here, and I can attach the file.

Current forum limitations do not allow for that size to be attached to posts, I can attach it to post#1 with summary of conclusions, if you lift the attachment limit.

Did you ever bother to check in another os install to see if it happens there? Completely ignoring your wd black.

I think that's a reasonable question. I rebooted into another OS, waited more than an hour and replicated the problem on both Windows 8.1 and Windows 10. Both patched to latest versions.


I share your suspicions Janus67, share them enough to realize that it's one thing for me to say what I want to say on an enthusiast forum here, but if I take it outside, I need a completely different person with another WD Gold like mine, to replicate this independently from me, before I can go elsewhere....


The other thing is, all my life using computers, I've encountered "what's a second or two... or three or four.... who cares?" attitudes.

This "issue" is relatively fast.
I don't think it is even full four seconds. I think it's less. But observable.
But fast enough to get a "what are you complaining about" objections from people..


Interestingly, that WD program they had me run does not work with WD Black drives. I just told them that again, after the Level 3 Engineer did reply again today:

This is a follow up to our call this afternoon on using wdckit. Per our conversation and while on the phone. First you ran wdckit S - to show all and it worked identifying that your WD Gold is on Disk 3. Also after running wdckit, and pulling the drive log, it appears that Idle_c is - 0 or not enabled so this would not be causing issues. We did talk about the WD gold and below you will find the Idle modes I have been sharing with you and we verified on the phone today.


Idle A - within 2 seconds, heads on media , full spin. Default Timer 20 (2 seconds)
Idle B , unloads heads, full spin at 10 min, Default Timer 6000 (10 Minutes)
Idle C, Disable, Default Timer NO


THIS ↑ is an excellent quote to illustrate the extent of support you can expect to get from Western Digital, you will get a tested determination of your drive's settings along with what modes mean. You will not get anything else.

I specifically said, "Okay I understand that HEADS ARE UNLOADED AFTER 10 MINUTES...

What does that actually mean, does it take a couple of seconds or more for the drive to go from HEADS ARE UNLOADED to fully operational state where everything comes up straight away?"

I actually asked him this, in that way. The answer was "I don't know".


What else could anyone expect from WD?
Well that depends on if another soul is reading this out there and has a WD Gold like mine.
If he or she cannot replicate this, then the entire case is an embarrassment for me over wasting so much time on what is a unique problem to my system.

If every WD Gold like mine does this, then surely someone at Western Digital organization is aware of this performance behavior, but a choice was made not to talk about it.

 

[Janus67]

I generally don't like recommending posting elsewhere but you could try the DataHoarders or homelab subreddits.

Did you ever hear back from anyone at Storage Review? They also have a subreddit as well.

 

[EarthDog]

Then surely someone at Western Digital organization is aware of this performance behavior, but a choice was made not to talk about it.

The Engineers.

That said, it is the intended behavior of the drive so, I'm not really sure it's worth it (for WD) to escalate. From their perspective, they seem to be making sure the drive is doing what it is intended to do. If I'm understanding correctly, it is.

I need a completely different person with another WD Gold like mine, to replicate this independently from me, before I can go elsewhere....

I think much earlier Janus suggested reaching out to Storagereview (or w/e) who reviewed a Gold. I think it's worth an email to them to ask if they experienced this or if they would kindly throw it on their test machine and see if it is repeatable for them. I'm sure if you explained things (executive summary...) to them, you may get a bite. It's likely nothing to do this for a reviewer with test systems, etc.

Did you ever hear back from anyone at Storage Review?

Poof! This ^^.

 

[freakdiablo]

The Engineers.

Ok, time for me to chime in.

The term "engineer" is thrown around way too much these days. It can literally mean anything. No BS, I was a "systems engineer working on international projects involving advanced image recognition & machine learning, radio transmital & ID codes, laser classification, and advanced data compilation techniques." Code for I crimped RJ45 connectors and knew how to configure a RAID controller, and I did that s*** for 7 years, but it looks good on paper (meaning my resume).

I get the sense that WD is doing the same here. They don't have a clue what's going on, so they're pointing fingers.

 

[c627627]

They wanted a screenshot to see why WD Black does not work under WDCKIT program they sent.

I sent one with a final push to get an answer.


wdckit_ WD2003FZEX.jpg shows the screenshot.

I understand that we established that for my WD Gold: HEADS ARE UNLOADED AFTER 10 MINUTES.

My question is if that means that after ten minutes, it takes two to four seconds for my WD Gold to go from HEADS ARE UNLOADED state to fully operational state where everything comes up straight away.
If that is so, then my entire case can be answered by concluding that this is the intended behavior of my WD Gold drive, so that's why that question is important.


View attachment 361367

 

[EarthDog]

Ok, time for me to chime in.

The term "engineer" is thrown around way too much these days. It can literally mean anything. No BS, I was a "systems engineer working on international projects involving advanced image recognition & machine learning, radio transmital & ID codes, laser classification, and advanced data compilation techniques." Code for I crimped RJ45 connectors and knew how to configure a RAID controller, and I did that s*** for 7 years, but it looks good on paper (meaning my resume).

I get the sense that WD is doing the same here. They don't have a clue what's going on, so they're pointing fingers.

Sure... I dig that.

But someone knows (job title be damned), is the point. Why they aren't getting to the bottom of it, no clue.

 

[freakdiablo]

Sure... I dig that.

But someone knows (job title be damned), is the point. Why they aren't getting to the bottom of it, no clue.

My money's on the CS department simply can't.

In all honesty there may be one person, maybe a small team in WD that worked on Gold spin down times. The CS department doesn't have a way to contact them. They're just going off some design documentation that person/team put together.

 

[c627627]

Everything I personally experienced points in this ↑ direction.

Next week when I get a reply will be a perfect opportunity to confirm it, because I asked the Level 3 Engineer yesterday,
"...we established that for my WD Gold: HEADS ARE UNLOADED AFTER 10 MINUTES.

My question is if that means that after ten minutes, it takes two to four seconds for my WD Gold to go from HEADS ARE UNLOADED state to fully operational state where everything comes up straight away."


If he says "I don't know" again, or if there is no answer, then what else are we supposed to conclude except that the CS department is just going off some design documentation that was put together by someone they have no way to contact.

 

[EarthDog]

My money's on the CS department simply can't.

In all honesty there may be one person, maybe a small team in WD that worked on Gold spin down times. The CS department doesn't have a way to contact them. They're just going off some design documentation that person/team put together.

Exactly. But an 'engineer' knows. They're just not going to spend their time digging deeper on intended behavior. If it wasn't working as intended, I bet they'd be able to contact them or further escalate the issue.

My question is if that means that after ten minutes, it takes two to four seconds for my WD Gold to go from HEADS ARE UNLOADED state to fully operational state where everything comes up straight away."

Didn't you already test and confirm there weren't any delays until 60 mins?

 

[c627627]

Level 3 WD Engineer answered today:

"Unfortunately, we cannot process your request or provide an ETA at this time.
We apologize for any inconvenience this has caused."

It was signed by the Level 3 engineer, who told me THERE IS NO higher level of support at Western Digital.

================================
So they don't know (!)

Let's review.

Western Digital first said:

"From the description you provided, you mention “a couple of seconds” this sounds like it’s in Idle_C and not actually spun down or in Sleep/Standby. Coming out of this idle state only takes a few seconds to get back to full speed and then load the heads. If the drive was spun down or in sleep/standby it could take up to 30 seconds to get to speed. Black does not support Idle_C which is why you see the difference. Unfortunately we do not provide a way of preventing the drive from going into this idle state."

THIS ↑ is how and why I thought my WD GOLD is going into IDLE_C ↑


Western Digital then provided software that shows that my WD Gold's firmware does NOT instruct my drive to go into IDLE_C, ever.

It showed that WD Gold goes into IDLE_A after two seconds and IDLE_B after 10 minutes.
Western Digital only shows those two states. Of interest is IDLE_B where WD says Heads are UNLOADED.


It therefore stands to reason that IDLE_B is the only suspect, according to Western Digital.

So then I boiled this ENTIRE thing down to this question I asked the Level 3 Engineer:

I understand that we established that for my WD Gold: HEADS ARE UNLOADED AFTER 10 MINUTES.
My question is if that means that after ten minutes, it takes two to four seconds for my WD Gold to go from HEADS ARE UNLOADED state to fully operational state where everything comes up straight away.


Western Digital official answer today is:
"Unfortunately, we cannot process your request or provide an ETA at this time.
We apologize for any inconvenience this has caused."


Therefore , not one person at Western Digital Support OFFICIALLY knows what the transition time in seconds is between the operational states of their hard drive.

 

[Janus67]

Whether they know or not isn't what is stated there. They basically gave you a non-answer/refused to answer the question and told you to not hold your breath to expect one.

 

[c627627]

I am reading your words, and they lead into speculation that they are knowingly hiding basic operational features of a very expensive drive.

 

[Janus67]

I guess I don't know what else you are expecting, beyond them confirming everything you've already found via the other conversations about idlec and inactivity after an hour causing a 3-4 second response time? Which on a running raid/nas (in I imagine the designed purpose solution) I imagine the drives will work together to not have a response time hit.

Ever hear anything from StorageReviews regarding their units?

 

[c627627]

I imagine the drives will work together to not have a response time hit.

Ever hear anything from StorageReviews regarding their units?

Hmmmm. What if you're right? What if this is only happening on non-raid, individual drive setups?

I have not contacted third parties.


If I were to buy again, I would buy Red.
But then why wouldn't I buy Seagate, given that there was zero warning from WD, or information about the "response time hits" anywhere at WD, even from their top-level support.

I remind you that all this, everything is speculation. Are the drives really doing "what they are supposed to" be doing as single drives? There is no info anywhere on this, from anyone.
If the drives cannot, or rather should not be used as single, non-RAID, regular drives, shouldn't Western Digital say that?


I was hoping someone reading this would chime in and say hey I just got a recent WD Gold and it is or it isn't happening to me.
If I could get just one person to confirm, it would dispel my fears that it might me my system.

But it's one thing for a system to wake up a sleeping / parked drive.
It's another thing for the system to actually put it to sleep / park it / unload it. Windows has one setting for turning off hard disks. It's not doing it on my system, so soon after last activity.

 

[c627627]

I opened this ticket with Western Digital on 14th of February 2023.
After continuous back and forth with their highest level of tech support, today May 2nd, 2023, I got this answer:


I apologize for the continued delay.

When the heads are on the ramp and the spindle is at lower speed, the yes, ~2 seconds is reasonable to come back to speed, load the heads and begin track following.

Thank you for your patience and apologize for the delays.


So, what does this mean, it sounds like they are confirming that WD Gold has a fundamental difference in functional operation from their other lines?

It means you are buying a performance crippled product - because you do not have to wait ~2 seconds with other lines, I mean that's what they are saying essentially?

 

[Zerileous]

You might need to refresh my memory, "heads on the ramp and spindle at lower speed" sounds like Idle_C again. So if you're looking at just heads unloaded without slowing down the spindle, that's Idle_B, which is supposed to happen after 10 minutes. But IIRC you tested and the problem doesn't appear until an hour out.

Going all the way back to post 22 https://www.overclockers.com/forums...tes-after-last-use.803327/page-2#post-8191217 it does state Idle_B to Idle_A has a usual time of 1s and a max of 30s. Perhaps it would take more testing to determine if this is just an unusual behavior with Idle_B. Also I note that Idle_C has an alternative name of Standby_Y, maybe that has caused confusion and perhaps the setting is listed somewhere else in their program. The behavior you've described is more consistent with Idle_C but is within the parameters of Idle_B, so more testing might be required to find the conditions which reproduce it.

All that said, I do have to play devil's advocate at some point. You assert that time is so important to you regarding a delay of a few seconds, so I'll ask a couple of questions. At what point does chasing down this rabbit hole take more time than you will spend waiting for the drive to respond over the course of the lifespan of the drive? I feel like you've already passed this point.

Is it correct that you have not followed the advice to reach out to other parties that may be able to comment on this issue, test if it is isolated to your drive or present in all golds, and if so then compare against reds, and publish the results for others to benefit? If so why continue to go circles with WD support who have given what they can, instead of trying an approach that may have more likelihood of success? A storage reviewer also may have a different channel to approach the company through that one could never access via public support channels.

It appears you're learning the difficult way that much of the purpose of a "customer support" program is to waste someone's time until they give up. This certainly is not limited to WD or even tech, most customer service is contracted out these days anyway, and they are set up to address common issues, but also to keep disgruntled customers from succeeding in occupying the time of an employee who has other responsibilities. Financially, it's inefficient for someone who actually knows the product well to spend time answering the question of someone with a single unit deployment that is outside of the designed use application. I'm sure if someone from AWS or Google had the same question they could have a legit answer in an hour, but you'll likely just keep getting apologies and canned spec's.

 

[c627627]

Thank you for your post.
I have already concluded that I will eventually use this drive as on off site back up drive only, because its life expectancy is longer than most, if not all other drives, and the delay when backing up would be irrelevant.
So when I see a good deal on a huge capacity drive - WD Gold will be replaced on my system.

I have already concluded that I will warn/ask anyone I ever come across who has this line of drives about their functionality.


It is correct that I have not reached out to Storage Review. I was hoping that at least one person at ocforums would eventually confirm my experience, at which point, I would reach out to third parties.
At some point I will - but my thinking of why I hadn't yet was based on self-doubt about my personal one experience with this drive.


You have analyzed the situation well, and I commend you on your great post - at some point in the future I will resurrect this thread with replies from third parties.