On Saturday, 15 June 2019 11:54:13 BST Ralph Corderoy wrote:
> It's probably correct, though you may want to double check or alter your
> means of connecting the drive to make sure it's not that causing
> problems. Or switch to another drive and hope badblocks finds no
> problems with that one, proving the connection's integrity.
I don't really have another (known good) drive available and I can't connect
this drive in any other way conveniently.
It is worrying that out of three 1 TB drives, two appear to be faulty, but
Hamish has the other two ATM. I came across a bug posted about three years
ago about smartctl throwing up false read errors, but I'm assuming the fix has
filtered through (although possibly not on the D-Link NAS box which has
horrendously out of date software on it).
> > I'm aware that the OS will work around badblocks
>
> Not really. A drive will remap sectors that it has trouble reading to a
> new sector when it either successfully manages to read its data, or is
> told by the computer to write new data to that sector. Until then, it
> reports a read error to the computer that will ripple up through the
> filesystem code in the kernel to the program and, if the programmer
> bothered to check for errors, the user.
>
> Separately, a filesystem, e.g. ext4, may have logic to track the
> locations of bad blocks and avoid them, e.g. mke2fs(8)'s -c and -l
> options. This is less useful AFAICS now a drive can re-map them itself,
> unless the drive is running out of new sectors from its reserves.
I suppose I meant the filesystem rather than the OS.
> > so the drive is probably usable for a time at least.
>
> I doubt it. But it may be educational to try.
We don't want to try with all our River System data...
> > We won't need all of the 1 TB of space for our purposes, so could we
> > partition the least bad part of the disc and use it?
>
> If the errors are constrained to a clear part to avoid.
There were just over 1000 badblocks reported, so not a huge amount in
comparison to a 1 TB drive.
> > If we did, what would happen when the NAS tried to build the RAID?
>
> No idea. Is it a mirror of the drives, or selected partitions on the
> drive?
Being a D-Link device there is zero support and very little information. I
presume it would try to mirror the whole thing, since that is what is would
expect to do.
Unless we can prove that these errors are false for some reason, then we need
to get our hands on another drive.
--
Terry Coles
--
Next meeting: BEC, Bournemouth, Tuesday, 2019-07-02 20:00
Check to whom you are replying
Meetings, mailing list, IRC, ... http://dorset.lug.org.uk/
New thread, don't hijack: mailto:dorset@mailman.lug.org.uk
