ext3 file system I/O blocks until reboot

[Robert Davidson]

Bruno Wolff III wrote:
> I saw stuff like this happening starting with later 2.6.20 kernels that
> wasn't fixed until the 2.6.24 kernels. (See bug 235043.) I wasn't using
> VM's, so it might not be the same as the bug you are seeing. I do remember
> seeing some other similar problems people were having that didn't appear
> to be the same bug as I had when I did bugzilla searches. So you might
> want to do your own bugzilla search to see what you can find.
>
> I have also been getting disk IO lockups in F10, but in a more limited set
> of circumstances. (Memory pressure on an X86_64 system.)
>   

Hi Bruno,

I've had a look through bugzilla but couldn't find any similar bugs (the
closest I can find is 439548 but I doubt very much that thats it).  Your
bug 235043 does sound rather different since it sounds like new
processes would be able to access the file system without a problem,
where as on my system any new attempt to read (writing wasn't tested)
just resulted in one more process stuck in the "D" state.

I might try taking a byte-for-byte copy of the FS and see if I can find
a way to reliably re-produce the problem on a similar server.

-- 
Regards,
Robert Davidson.
Obsidian Consulting Group.
Ph. 03-9355-7844
E-Mail: support at obsidian.com.au