[arch-general] [OT]Disk showing too many bad sectors - is it going to fail ?

Mauro Santos registo.mailling at gmail.com
Sun Oct 17 12:46:07 EDT 2010


On 17-10-2010 15:55, Partha Chowdhury wrote:
> i overwrote the whole disk with ddrescue -f /dev/zero /dev/sdb.After one
> and a half hours later it stopped with the message "no space left on
> device" - i guess it indicates no problem ?

> i also tried the badblocks program with -w option. It took a long time
> 5+ hours but did not report a bad sector.

It seems to indicate that everything is ok

> I checked the seagate site and there is no firmware upgrade for this
> model. On further googling, i found that seagate is only offering
> firmware upgrades for 7200.12 model onwards.
> 
> Now to be absolutely sure, i downloaded the seatools program and it ran
> a short and long test which both said PASSED.

This reinforces the conclusion taken after the previous tests. If I'm
not wrong the test seatools performs is just issue a smart long test,
which you can also do with smartctl (and also check the test log), this
is a read only test as far as I know. In the smart test log you can also
see the addresses of the current pending sector count and offline
uncorrectable sectors if they are not too many and if the drive returns
sane data (which doesn't seem to be the case).

> Inspite of all these, gsmartcontrol shows the same.
> 
> What are the indications before a disk is going bad which a normal user
> can catch with bare eyes and ears ?

The first indication would be frequent hangups or trouble reading some
files and errors on dmesg about not being able to read sector xyz and
some (s)ata code error. However like I said before, I've experienced a
case where a write to the affected sectors solved the problem, but if
the problem is serious then at best the sector will be marked as damaged
and reallocated if there is still spare space for reallocation.

If you start to see the reallocated sector count (and reallocation event
count on some drives) increase then better backup everything.

Also keep an eye on all pre-failure attributes, if any of those says
failing or failed in the past I think it's bad news.

You may also hear some abnormal clicking, however if you hear this the
drive is probably already way past any possibility of data recovery at home.

Mind you that the sometimes drives fail without warning or without any
change in the smart attributes. This [1] is an interesting read.

[1] http://labs.google.com/papers/disk_failures.pdf

-- 
Mauro Santos


More information about the arch-general mailing list