Lance:
Thanks a lot for explaining this!
I'm still trying to outrule potential causes to the
crashes I reported earlier; one by one.
Right now I'm investigating if swap on RAID might be the culprit.
Thanks again to everybody who answered with suggestions and
advice, I've got stuff here to be experimenting for a week or
more.
--
An encouraging side-note (totally off the subject):
Sweden's national phone company, Telia, has a very famous and
popular service called "Miss Clock". "She" tells you the correct
time in a gentle voice. This service has been provided
for some 50 years or more. The other day they upgraded
the system, switching from old mechanical tape recorders to
a totally computerized, atom-clock-driven system. Guess what
OS they based their new solution on - Linux.
I was just so happy hearing about it that I had to share it with
someone, and that happened to be you. :)
Best regards,
/Johan Ekenberg
> This is what is happening...
>
> 1. You rebooted
> 2. The raid volume is started, and since it isn't clean (bad shutdown)
> it starts resyncing at the default 1024 byte buffer size
> 3. Your filesystem is mounted, and it seems to be using 4096 byte blocks.
> This changes the buffer block sizes to 4096.
> 4. The resync process was in the middle of doing 1024 byte transfers
> when the block size was changed, thus, you get all the
> error messages
> as noted.
> 6. The resync process notices the block size change and re-issues
> the request.
> 7. Eventually the 1K buffers get too old and are reclaimed for something
> else (I think).
>
> This is not a problem, just a bunch of messages.
>
> <>< Lance.
>
>
> > After crashes I see a lot of these messages (RAID5):
> >
> > www3 kernel: ll_rw_block: device 09:00: only 4096-char blocks
> implemented
> > (1024)
> > www3 last message repeated 226 times
> > www3 kernel: md0: blocksize changed during write
> >
> > What do they actually mean?
> > The first one (only 4096-char blocks implemented)
> > totally fills my dmesg output.
> >
> > /Johan Ekenberg
> >
> >
>
