Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.sys.ibm.pc.hardware.storage > #8596 > unrolled thread

Interpreting diskscan results

Started bybp@www.zefox.net
First post2026-10-28 15:24 +0000
Last post2026-11-14 01:54 +0000
Articles 5 — 3 participants

Back to article view | Back to comp.sys.ibm.pc.hardware.storage


Contents

  Interpreting diskscan results bp@www.zefox.net - 2026-10-28 15:24 +0000
    Re: Interpreting diskscan results Jim Jackson <jj@franjam.org.uk> - 2026-11-03 20:23 +0000
      Re: Interpreting diskscan results bp@www.zefox.net - 2026-11-04 14:57 +0000
        Re: Interpreting diskscan results not@telling.you.invalid (Computer Nerd Kev) - 2026-11-13 06:49 +1000
          Re: Interpreting diskscan results bp@www.zefox.net - 2026-11-14 01:54 +0000

#8596 — Interpreting diskscan results

Frombp@www.zefox.net
Date2026-10-28 15:24 +0000
SubjectInterpreting diskscan results
Message-ID<11bt43t$2gksl$1@dont-email.me>
A Raspberry Pi 2B self-hosting FreeBSD-current developed what
looks like filesystem corruption. Most correspondents suggest
that something's wrong with the disk. The only non-destructive
test I could find was diskscan, a Linux program available on
RasPiOS, which seems to plot read access time vs number of 
blocks having that access time. Nothing hugely amiss is apparent,
but I'm no expert with diskscan. 

Five scans have been run, three with the disk directly connected
to a Pi5 and two using a Pi2B connected via the same powered hub 
& supply that were used when the corruption developed under FreeBSD.

The diskscan output is at
http://www.zefox.net/~fbsd/rpi2/disktrouble/
If anybody can offer insights into what the scan results mean
it would be much appreciated.

If there is better software to use for testing, please indicate.
I'm aware of diskroaster on FreeBSD, but that's obligately destructive.

Thanks for reading,

bob prohaska
 

[toc] | [next] | [standalone]


#8603

FromJim Jackson <jj@franjam.org.uk>
Date2026-11-03 20:23 +0000
Message-ID<slrn11ekgm2.2dg.jj@iridium.wf32df>
In reply to#8596
On 2026-10-28, bp@www.zefox.net <bp@www.zefox.net> wrote:
> A Raspberry Pi 2B self-hosting FreeBSD-current developed what
> looks like filesystem corruption. Most correspondents suggest
> that something's wrong with the disk. The only non-destructive
> test I could find was diskscan, a Linux program available on
> RasPiOS, which seems to plot read access time vs number of 
> blocks having that access time. Nothing hugely amiss is apparent,
> but I'm no expert with diskscan. 

is smartctl or equiv. available on FreeBSD? You need to try and check 
the S.M.A.R.T. stats that disks keep nowadays. You may need to 
do a web search.

>
> Five scans have been run, three with the disk directly connected
> to a Pi5 and two using a Pi2B connected via the same powered hub 
> & supply that were used when the corruption developed under FreeBSD.
>
> The diskscan output is at
> http://www.zefox.net/~fbsd/rpi2/disktrouble/
> If anybody can offer insights into what the scan results mean
> it would be much appreciated.
>
> If there is better software to use for testing, please indicate.
> I'm aware of diskroaster on FreeBSD, but that's obligately destructive.
>
> Thanks for reading,
>
> bob prohaska
>  
>
>

[toc] | [prev] | [next] | [standalone]


#8605

Frombp@www.zefox.net
Date2026-11-04 14:57 +0000
Message-ID<11cfh4e$oj14$1@dont-email.me>
In reply to#8603
Jim Jackson <jj@franjam.org.uk> wrote:
> On 2026-10-28, bp@www.zefox.net <bp@www.zefox.net> wrote:
>> A Raspberry Pi 2B self-hosting FreeBSD-current developed what
>> looks like filesystem corruption. Most correspondents suggest
>> that something's wrong with the disk. The only non-destructive
>> test I could find was diskscan, a Linux program available on
>> RasPiOS, which seems to plot read access time vs number of 
>> blocks having that access time. Nothing hugely amiss is apparent,
>> but I'm no expert with diskscan. 
> 
> is smartctl or equiv. available on FreeBSD? You need to try and check 
> the S.M.A.R.T. stats that disks keep nowadays. You may need to 
> do a web search.
> 

Yes, smartmontools are available on FreeBSD. The first few attempts
reported SMART wasn't supported, but eventually a response came back.
It's at
http://www.zefox.net/~fbsd/rpi2/disktrouble/smartctl.log

A correspondent on the FreeBSD mailing lists pointed out a few
bad omens:
  1 Raw_Read_Error_Rate     0x000f   083   056   006    Pre-fail  Always       -       196488790

  5 Reallocated_Sector_Ct   0x0033   099   099   036    Pre-fail  Always       -       386

  9 Power_On_Hours          0x0032   069   069   000    Old_age   Always       -       27847 (91 61 0)

I instead paid attention to 

197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always       -       0

and didn't consider three years runtime excessively long.

Evidently, I was mistaken 8-) , thinking there'd be something
like "spare sectors exhausted" or some other explicit message
that the disk was EOL. 

Somewhat curiously, both diskscan and dd if=/dev/da1 of=/dev/null
seemed to indicate no explicit errors. Diskroaster, a FreeBSD port,
reported five errors.

Thanks for writing,

bob prohaska

 

[toc] | [prev] | [next] | [standalone]


#8609

Fromnot@telling.you.invalid (Computer Nerd Kev)
Date2026-11-13 06:49 +1000
Message-ID<6af62676@news.ausics.net>
In reply to#8605
bp@www.zefox.net wrote:
> Evidently, I was mistaken 8-) , thinking there'd be something
> like "spare sectors exhausted" or some other explicit message
> that the disk was EOL. 
> 
> Somewhat curiously, both diskscan and dd if=/dev/da1 of=/dev/null
> seemed to indicate no explicit errors. Diskroaster, a FreeBSD port,
> reported five errors.

A bit late, but for a more conclusive go/no-go answer from SMART
I'd run a self-test and check whether it fails:

smartctl -t long /dev/da

It'll say when it expects to finish, after which time you can check
the result with (ignoring the old self-test logs with earlier
lifetime hours):

smartctl -l selftest /dev/da

Or I suggest running that every so often before the extimated
completion time to check if it's failed early with errors (note
that it might not show the new test in the log until it's
completed/failed). From the looks of things, I'd expect that
drive to fail with an error fairly quickly.

-- 
__          __
#_ < |\| |< _#

[toc] | [prev] | [next] | [standalone]


#8610

Frombp@www.zefox.net
Date2026-11-14 01:54 +0000
Message-ID<11d8f0c$1ieti$1@dont-email.me>
In reply to#8609
Computer Nerd Kev <not@telling.you.invalid> wrote:
> bp@www.zefox.net wrote:
>> Evidently, I was mistaken 8-) , thinking there'd be something
>> like "spare sectors exhausted" or some other explicit message
>> that the disk was EOL. 
>> 
>> Somewhat curiously, both diskscan and dd if=/dev/da1 of=/dev/null
>> seemed to indicate no explicit errors. Diskroaster, a FreeBSD port,
>> reported five errors.
> 
> A bit late, but for a more conclusive go/no-go answer from SMART
> I'd run a self-test and check whether it fails:
> 
> smartctl -t long /dev/da
> 
> It'll say when it expects to finish, after which time you can check
> the result with (ignoring the old self-test logs with earlier
> lifetime hours):
> 
> smartctl -l selftest /dev/da
> 
> Or I suggest running that every so often before the extimated
> completion time to check if it's failed early with errors (note
> that it might not show the new test in the log until it's
> completed/failed). From the looks of things, I'd expect that
> drive to fail with an error fairly quickly.
> 

In the end I could not get a clear failure indication from either
diskscan or SMART self-testing. Eventually it developed that the
filesystem wouldn't stay clean, even after running fsck -fy. 

I tried diskroaster, a destructive test in the FreeBSD ports
collection. It clearly reported four bad sectors. I imagine
that suggests the problem is with writing the disk, not reading
it. 

Is there a way to get the disk to do a destructive self-test
that reassigns the unwritable blocks, analgous to an old-fashioned
low-level format? I'd  not want to use the disk for anything
important but it would be an interesting experiment if it can
be done. I vaguely recall that low-level formatting can't, or
at least shouldn't, be done outside the factory on an LBA 
device, but there's nothing much to lose at this point.

Thanks for writing!

bob prohaska
 

[toc] | [prev] | [standalone]


Back to top | Article view | comp.sys.ibm.pc.hardware.storage


csiph-web