On Wed, 2007-05-16 at 21:10 +0300, Itay wrote:
> On Tue, 15 May 2007, Richard Karhuse wrote:
>
> > <snip headers>
> ><snip>
> ** I failed to mention earlier that the media check prior to
> installation *failed*. I decided to go on with the installation
> because my own experience so far, and others', showed that
> failure does not necessarily imply bad media. But see below.
Based on investigation I've done in response to the thread here,
http://lists.centos.org/pipermail/centos/2007-April/079718.html
and the advice by several in that thread, you need to have padding on
the media. I'm authoring a "SOLVED" message for that thread, but haven't
completed my tests.
Preliminary conclusions are that the Linux read-ahead feature has a
"bug". If the read is near end-of media and the system is filling
buffers, it seems that it is ignorant of the fact that the file system
size (written media size in this case) has been exceeded. This has been
reported as affecting ISO-9660 file systems, but I can testify that it
also affects "raw" operations such as "dd".
I *suspect* that speed of equipment (burner, memory, PCI bus, CPU,...)
affect this, but I've not yet tested on two slower nodes to confirm
this. Real life keeps interfering.
In association with those error messages, you might see some surrounding
errors mentioning the "logical blocks" involved with the errors. If
those logical blocks are *within* your written media size, you have a
definite problem with the media or hardware, IMO.
If they are beyond the written media, it's the kernel/driver error, IMO.
Is it a problem? Depends, I think. For things such as "dd", no. It
finishes writing what it could and all is well in la-la land.
For other applications, if the error is reported while they are still
processing valid data, the application(s) should/could terminate without
completing processing.
Then you have problem.
Use one of the padding methods mentioned in that thread and you should
be OK and the errors should disappear.
> <snip>
> anaconda.syslog seem to be more interesting
> ===========================================
>
> There are quite a few
>
> <4>hda: media error (bad sector): status=0x51 { DriveReady SeekComplete Error }
> <4>hda: media error (bad sector): error=0x30 { LastFailedSense=0x03 }
As mentioned above, ...
>
> Followed by
>
> <3>Buffer I/O error on device hda, logical block ...
>
> (hda is my DVD drive, so I guess it is related to the failed
> media check above?)
>
> <snip>
> (Also related to the failed media check above?)
My *guess* is that the application related errors you reported may be a
result of certain installation steps terminating early due to the false
I/O errors reported by the kernel/driver(s).
> <snip>
HTH
--
Bill
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
