Hi Mauro,

Thanks for this hint. But my script looked in the beginig like yours :-)

I have tried it out.

CHCCWDEV -d gives sometime rc 256.
and the fdasd rc: 65280 (Disk in use!).

I added now retries: In case I have a nonzero rc I redo the command up to
five times with a sleep of one second.
Up to know the retry loop has to issue the commands only a second time to
deliver an rc 0.

This is what I don't like: "Hammering" the command into the system and hope
it will work out. I used sync which might help for the fdasd problem after
the dasdfmt. But I am nut sure yet.

BTW: I have kernel 3.0.34-0.7-default active on my system.

BR
Florian



On Mon, Jul 16, 2012 at 9:54 PM, Mauro Souza <[email protected]> wrote:

> I usually put some sleeps and syncs along the way to give time to this
> settle down:
> dasdfmt -y -b 4096 -p -f /dev/dasdx
> sleep 2
> sync
> fdasd -a /dev/dasdx
> sleep 2
> sync
> mkfs.ext3 /dev/dasdx1
>
> It works almost all the time. You could put some error checking routines
> around, but I think this is simple enough to run without that.
> Em 16/07/2012 15:45, "Florian Bilek" <[email protected]> escreveu:
>
> > Dear all,
> >
> > From time to time I am confronted with an weird problem with the
> s390-tools
> > which is really a PITA.
> >
> > I have a small script that should simply execute a dasdfmt afterwards a
> > fdasd and finally a mkfs to prepare a new DASD.
> >
> > What happens is that after the dasdmft the fdasd gets sometimes the
> > information that the disk is in use and gives an return code and the
> script
> > stops.
> >
> > This is results now in a kind of game to serialize those commands.
> >
> > In SLES 11 SP1 I could find a solution that after the dasdfmt I would
> > disable the device (chccwdev -d xxxx) and enable it afterwards. There
> was a
> > wait of some seconds in between.
> >
> > Looking on the Internet I found a better way to circumvent the fixed
> > waiting time: udevadm settle --timeout=60 --exit-if-exists=/dev/dasd
> would
> > wait till either there is a timeout or the device shows up in the /dev
> > filesystem.
> >
> > Under SP2 I am facing again the problem since either a chccwdev -d xxxx
> > brings now a device busy and delivers a return code or the fdasd returns
> > that the device is busy.
> >
> > Is there a way to safely check if the device is not busy any more?
> >
> > That device is LINKED with W exclusively to the guest executing my
> script.
> > There is also no other activity on that specific device on this guest.
> >
> > I have the feeling that a new kernel brings sometimes back these
> troubles.
> >
> > Is there a proper way to address this problem?
> >
> > Thank you very much in advance.
> >
> > --
> > Best regards
> >
> > Florian Bilek
> >
> > ----------------------------------------------------------------------
> > For LINUX-390 subscribe / signoff / archive access instructions,
> > send email to [email protected] with the message: INFO LINUX-390 or
> > visit
> > http://www.marist.edu/htbin/wlvindex?LINUX-390
> > ----------------------------------------------------------------------
> > For more information on Linux on System z, visit
> > http://wiki.linuxvm.org/
> >
>
> ----------------------------------------------------------------------
> For LINUX-390 subscribe / signoff / archive access instructions,
> send email to [email protected] with the message: INFO LINUX-390 or
> visit
> http://www.marist.edu/htbin/wlvindex?LINUX-390
> ----------------------------------------------------------------------
> For more information on Linux on System z, visit
> http://wiki.linuxvm.org/
>

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to