On 2020-09-04 05:38, Simon Walter wrote:
> On 9/4/20 3:50 PM, golinux@??? wrote:
>> Well . . . I decided to run an fsck on the misbehaving harddrive. It
>> started off by identifying the errors and rewriting them and then went
>> through Free block counts, Inode bitmap differences and Free inodes
>> and directory count. Some snippets of the output are posted below. I
>> did not stick around to watch all of it so may have missed something.
>> It took a long time to sort itself!
>
> Out of curiosity, was it connected via eSATA or USB? I forgot what
> kind of dock you got.
>

The drive is still in the external case. Connection is USB with AC
adapter. Testing it on the dock would have been the next step. I did
finally get the dock out of the box but have not fired it up yet. I am
old, slow and methodical . . .

>>
>> When it finished, I mounted the drive without issue and could read the
>> remaining directories and files.  However the /media/xxxxxx/cstwo/600
>> directory mentioned in the original mounting error below was nowhere
>> to be found:
>>
>> "Error when getting information for file '/media/xxxxxx/cstwo/600':
>> Input/output error."
>
> It's possible that data is not actually gone and can be recovered if
> you know what you are looking for. However, as you stated, you have a
> copy of the data elsewhere.
>

A few posts after yours Hendrik suggested checking the lost and found
directory and in it I found 57 folders and 15798 other items totaling
16.6 GB. They are mostly edits from audacity (.au) and avidemux (in C).
Also some wav and mp3 and graphics stuff too. I couldn't find the
associated mpgs or isos and there seems to be some other things missing
here and there.

Everything in the lost and found can be deleted. Maybe if I do it in
small chunks, it won't explode. LOL!

>>
>> Then I ran the SMART Short offline test and it completed without error
>>
>> I have no idea how reliable the repaired drive is after this radical
>> surgery. Can it be written to or files deleted? Should I even try?
>
> From my experience, SMART data is reliable. To give some idea, I have
> experience with over a hundred disks since they started to include
> SMART. So probably not that much compared to others on this list.
>
> I would say there is no problem with the underlying disk and the
> corruption occurred at the filesystem level, which is one reason an
> entire directory is missing, rather than at the disk level. If you
> want more assurance, run the long test. You can get some idea of how
> quickly your drive is deteriorating by monitoring changes to the SMART
> data (smartd). I've been able to predict failure before it happens.
> It's never been sudden. So if your disk "PASSED" it's probably fine to
> use it.
>
> Modern disk drives will move your data to good sectors when it detects
> failure looming in bad sectors. So head failure is an issue, and can
> also be predicted by SMART data. Mishandling of drives is something
> that SMART can't predict of course. ;)
>

Thanks for the useful info Simon. I will run the long SMART test later
today.

> Best regards,
>
> Simon
>