Mailing List archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[vdr] Re: Maxtor 160 GB



Ralf Spitzauer wrote:
> 
> Carsten Koch wrote:
> ...
> > If my disk test below completes a full 153GB write and read pass
> > on your 153GB disk, you are all set with whatever you have.
> >
> 
> Hello Carsten,
> 
> afer building a new Kernel 2.4.16 and paching with ide.2.4.16.02152002.patch
> Yast shows me correct number of cylinder and correct capacity.
> But your disk_test reports an error:
> 
> linux:/vdr/disktest # ./disk_test /dev/hdg1
> Writing.
> 156327 MB.
> Write complete - Input/output error.
>  163921575936 bytes in 5706.72 seconds = 27.3935 MB/s.
> Reading.
> 156327 MB.
> Read aborted - Input/output error.
> 163921571840 bytes in 5703.06 seconds = 27.4111 MB/s.
> linux:/vdr/disktest #
> 
> What kind of problem is this, what can I do?

This is a bit strange.
First of all, both the write and the read section of the test 
run until an error occurs (or read gets an EOF), as the test 
does not know a maximum size. Under Linux, the system reports 
the (somewhat generic)  "Input/output error". 
For example, IRIX is more precise and reports "no space left on device".

So, it is expected to get an Input/output error at the end of 
the test. However, you got an Input/output error milliseconds
before the end of the test. 4K have been written but not read.
I do not know what could have caused this. Maybe someone on this
list, who has more disk driver knowledge can explain it?

Of course, it could really be the obvious reason (your
disk has a media error near the end). In fact, when I 
look at the results of a test that I ran on the same
type of disk:

vdr:/tmp # ./disk_test /dev/hdc
Writing.
156334 MB.
Writing complete - Input/output error.
 163928604672 bytes in 5650.2 seconds = 27.6688 MB/s.
Reading.
156334 MB.
Reading complete - Input/output error.
 163928604672 bytes in 5673.3 seconds = 27.5561 MB/s.

I can see that your test ends 7 MB earlier.
IIRC, I checked the size back then and found mine to
be exactly right, so your test must have ended prematurely
both during write and read!

I believe that the disk driver reports any errors it
encounters to /var/log/messages.
Did anything show up there while you ran the test?
If yes, you definitely have a bad disk.

Everything else looks good. Apparently you successfully 
wrote and read 152 GB, so 48-bit-LBA is definitely working.
I would not trust the disk, though, before it passes my
test at least two times in a row.
You may also want to run the badblocks test on that disk.



> Thanks a lot for helping

You are welcome.

I will be on vacation during the next week, so
if you have any more questions, I can answer
them tonight or in a week.

Carsten.



Home | Main Index | Thread Index