2

I'm running a desktop with Ubuntu 22.04.1 LTS, with the Mate desktop environment installed on top. I recently acquired a 10TB HDD from Seagate, part number ST10000NE000-3AP101, firmware version EN01.

The drive is unprecedentedly loud. Louder than any drive I've encountered in my 30 years of computer use (apart from those with actively crashed heads).

  • It is quite irritating and not really compatible with me focusing on work with this noise level. I definitely can't have it running when other people are in the shared office.
  • I'm reasonably concerned that Linux is rapidly destroying the new drive.

Usage pattern:

  • The specific sound is bursts of 8-20 Hz grinding noises, of variable duration, in irregular patterns. It is currently copying a lot of data, but there is no excuse for this noise level. It doesn't sound like the click of head parking, so my guess is that the head is seeking back and forth at this frequency, but with far too brutal a controller.
  • I'm not using it as a system disk, but I am currently backing up data to it.
  • The Western Digital drive that this drive replaced ran silently, even under identical heavy loads, so it's not an issue with the mounting/tower/etc.

Why this question is not (obviously) a duplicate

This is a pretty common problem, but I've spent all morning across various forums with no satisfaction. I'll collect what I've found here.

  • It's not the head-crashing sound. The drive seems to work fine, for now.

  • Manipulating the IO pattern is also not relevant—The identically mounted WD drive handled similar IO patterns silently.

  • Comments on these questions (1, 2, 3) indicate that these noises are absolutely abnormal for modern HDDs. I tend to agree: It's louder than the HDDs from the early 90s. Should I be returning the drive to the manufacturer?

  • Lot's of answers (e.g. 1,2,3) suggest changing acoustic or spindown parameters, but this doesn't work:

    φ sudo hdparm -M /dev/sdd
/dev/sdd:
 acoustic      = not supported
φ sudo hdparm -B /dev/sdd
/dev/sdd:
 APM_level  = not supported

  • This question might be the same issue, but I suspect not. iotop reveals that most of the disk load is from Dropbox (which makes sense, I'm sync'ing it now). However, as I mentioned earlier, the previous WD drive handled this same load silently, and I've never heard a drive this loud. This leads me to believe that the drive is experiencing excessive mechanical noise, even under normal usage patters, which will shorten its lifespan.

  • You'll find some answers saying noise is normal—but this not a normal sound. I'm inclined to take warnings that this will shorten drive lifespan seriously. From that link, this comment in particular rings familiar:

my problem is covered a lot online if you search [comments on Linux forums] say is the hard-drive manufacturer setting too aggressive power saving settings, but i don't believe this because it does not happen in windows and i refuse to believe that Seagate set their laptop drives in a self destruct mode, they would have lots of peed of customers on their hands and at risk of having licenses revoked...

  • It's also not this issue, since hdparm -I /dev/sdd works fine (and it is an ATA drive). There are some hints that failure of the -B and -M settings can be a fixable communication issue, but the aforlinked solutions are for a specific (different) case.

What I think is happening

I believe that this sound is, in fact, abnormal and indicates a problem. At best, the drive is running within design parameters but in some sort of "maximum speed, minimum lifespan" mode, which is absolutely not what I want.

hdparm -I thinks the drive has the following capabilities:

Capabilities:
    LBA, IORDY(can be disabled)
    Queue depth: 32
    Standby timer values: spec'd by Standard, no device specific minimum
    R/W multiple sector transfer: Max = 16  Current = 16
    Recommended acoustic management value: 254, current value: 0
    DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6 
         Cycle time: min=120ns recommended=120ns
    PIO: pio0 pio1 pio2 pio3 pio4 
         Cycle time: no flow control=120ns  IORDY flow control=120ns
Commands/features:
    Enabled Supported:
       *    SMART feature set
            Security Mode feature set
       *    Power Management feature set
       *    Write cache
       *    Look-ahead
       *    WRITE_BUFFER command
       *    READ_BUFFER command
       *    DOWNLOAD_MICROCODE
            Power-Up In Standby feature set
       *    SET_FEATURES required to spinup after power up
            SET_MAX security extension
       *    48-bit Address feature set
       *    Mandatory FLUSH_CACHE
       *    FLUSH_CACHE_EXT
       *    SMART error logging
       *    SMART self-test
       *    Media Card Pass-Through
       *    General Purpose Logging feature set
       *    WRITE_{DMA|MULTIPLE}_FUA_EXT
       *    64-bit World wide name
       *    IDLE_IMMEDIATE with UNLOAD
            Write-Read-Verify feature set
       *    WRITE_UNCORRECTABLE_EXT command
       *    {READ,WRITE}_DMA_EXT_GPL commands
       *    Segmented DOWNLOAD_MICROCODE
       *    unknown 119[6]
       *    unknown 119[7]
            unknown 119[8]
            unknown 119[9]
       *    Gen1 signaling speed (1.5Gb/s)
       *    Gen2 signaling speed (3.0Gb/s)
       *    Gen3 signaling speed (6.0Gb/s)
       *    Native Command Queueing (NCQ)
       *    Phy event counters
       *    Idle-Unload when NCQ is active
       *    READ_LOG_DMA_EXT equivalent to READ_LOG_EXT
       *    DMA Setup Auto-Activate optimization
            Device-initiated interface power management
       *    Software settings preservation
            unknown 78[7]
       *    SMART Command Transport (SCT) feature set
       *    SCT Write Same (AC2)
       *    SCT Error Recovery Control (AC3)
       *    SCT Features Control (AC4)
       *    SCT Data Tables (AC5)
            unknown 206[7]
            unknown 206[12] (vendor specific)
            unknown 206[14] (vendor specific)
       *    SANITIZE_ANTIFREEZE_LOCK_EXT command
       *    SANITIZE feature set
       *    OVERWRITE_EXT command
       *    Extended number of user addressable sectors

Thus, I think that what's happening is that the drive controller can be configured to reduce noise and wear (and is quite possibly not supposed to be using the current setting). However, It seems like Linux, or at least hdparm, does not know how to communicate with the drive to set these options. This would mean that there is some magic set of bytes I'd be able to send to the device to fix this issue, but it may require a bespoke solution specific to these Seagate drives?

Update:

After watching this video, it seems like this line of drives might just be designed to R/W fast and die young. I strongly recommend not buying the Seagate drives, except perhaps for a rack mounted system acoustically isolated in a separate room. I will still hold out hope that there is a fix (firmware update, bespoke commands), but the answer may simply be: don't buy these drives for your office desktop, or if you want something that will not vibrate itself apart.

Suggested concrete next steps:

  • I want to explore all options for reducing what I think is a noise related to rapid head-seeking. I'm willing to explore difficult solutions like finding new drive firmware, or writing some custom C code to get/set drive settings not supported in hdparm
  • First, let's solve this mystery: hdparm -I reports Recommended acoustic management value: 254, current value: 0; If the drive doesn't support acoustic management, where is this default coming from? Is it hard-coded in the hdparm program? (I could believe that the "current value" is nonsense, if no such parameter exists on the drive).
Improve this question
11
  • 1
    Voting as a duplicate doesn’t work like you want, if a user believes an existing answer actually answers your question, they should vote to your question. If the question is closed, you should edit your question, and with a detailed explanation, make it clear that existing answer does not answer your question. That edit if it becomes necessary shouldn’t just be a complaint your question was closed. Meta commentary like that exists in your question should be removed. As far as your question is concerned I assume just returning the drive isn’t an option
    – Ramhound
    Jul 25 at 11:51
  • Why all this effort? If you suspect the drive, copy the data onto a new one and return it.
    – Jan Doggen
    Jul 25 at 11:56
  • 1
    I'm at an institution+culture+country not known for efficiency. It took me multiple days, full time, to order the drive through our internal processes. We can only order through approved suppliers, which have no useful search features, inaccurate "in stock" information, higher prices, and slower turnaround. It took the drive two months to arrive. If there is a "hacky" solution that avoids interacting with procurement, it will be much faster. I don't quite have energy to go through the bureaucracy again today, and wait another N months. My contract ends soon anyway. Perhaps tomorrow.
    – MRule
    Jul 25 at 12:13
  • 1
    What I've learned is that in 2008, Seagate stopped paying for the patent that lets you control drive noise. I've been mostly using SSDs since then, so I probably didn't notice. I went back to HDD to get more storage/$. If you haven't used an HDD since the early 2000s, it turns out that they ARE worse these days---and perhaps even louder than 1990s drives, which were forced to operate more slowly due to design constraints. I'll close the question. Disappointed, but at least I know why now. Always a shame when technology regresses.
    – MRule
    Jul 25 at 13:38
  • 1
    @MRule - I actually never thought your question should be closed for not being within scope. I just was allowing you a chance to not make the same mistake as users in the past, or explain how to get it to be reopened, if it was eventually closed.
    – Ramhound
    Jul 25 at 14:20

1 Answer 1

Reset to default
1

After doing some reading (wiki), apparently Seagate stopped allowing you to control noise levels in 2008 because they did not want to pay for a patent.

If you've mainly used SSDs since then, you'll be surprised to find that modern HDDs have, at least in some ways, regressed since the 2000s. My memories of the 1990s is hazy, but it's possible drives were quieter then because we didn't have the materials to run them aggressively without causing failure. These new Seagate drives sound like an old relay computer and I imagine involve similar forces/jerk to a mid-sized mechanical relay.

The reason you might not have noticed this on, say, external HDDs is that while Seagate (and apparently WD now too) don't allow you to adjust noise levels, they are allowed to tune each product for a noise level related to its expected operational requirements. So, a slower drive or external drive intended for office use can be given fixed parameters that make it quieter.

These larger Seagate ("IronWolf") drives run at 7200 RPM, which may well be necessary if you hope to R/W many GB of data from to/from them. I have no doubt that it is possible to adjust the seek controller to reduce noise, and the drive firmware probably supports this. However, they aren't allowed to expose this feature to the consumer without paying for patents. Without the ability to tune drive parameters, the Seagate IronWolf series really only belong on high-performance rack-mounted servers, where you want speed at the expense of acoustics and perhaps longevity.

I would assume that much of Seagate drive firmware is shared across their various HDDs, and that they are fixing different speed/acoustics/longevity settings for each product's expected market. So it's probably possible to fix this—either through a trade-secret set of serial commands, or by patching the drive firmware. I'm not quite up to that level of hacking, but it is the sort of thing I'd expect to see on HackerNews one of these days. Until then, case closed.

FAQ

  • Yes, it's too loud for an office or home. Also, it affects Seagate drives in general, not just this model. Many others corroborate (1,2,3,4,5,).
  • No, you can't change it (patent reasons, not technology reasons); You aren't going crazy: In this sense, HDD drives from Seagate have regressed since the 2000s.
  • Yes, the drives are (probably) physically capable of fixing this—but someone will need to crack/reverse-engineer the firmware to do it.
  • No, you should not buy a Seagate IronWolf drive for an office computer. Use these as high-performance (expendable) drives in performance servers.

(You may recall similar feature regressions in sound cards in the mid 2000s, when legally-imposed anti-counterfeiting measures were implemented to make capturing the audio loopback more challenging)

Improve this answer
1
  • 1
    Older mechanical drives were a great deal less sophisticated. They have double and tripled the number of platters in a drive. We have multiple ways to write data to the disks and handle methods to increase the number of platters using Helium. My point the disks of yesterday were less complicated.
    – Ramhound
    Jul 25 at 14:19

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .