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

Re: [rhel6-branch] & [master] format unformatted DASDs (#528386)

Hash: SHA1

On Tue, 27 Oct 2009, Hans de Goede wrote:


On 10/27/2009 03:07 AM, David Cantrell wrote:
These patches are intended for rhel6-branch and master.

Bug #528386 is meant to improve the dasdfmt support we had from RHEL-5.  I
have divided the work in to two parts:

     1) Functional.  Detect unformatted DASDs, offer the user to option
        to format them or bail, perform the format and display progress.
     2) Parallel formatting.  IBM wants parallel formatting support for
        sites that have thousands of DASDs.  I wanted to separate that
        work from these patches.

I changed execWithPulseProgress() in iutil.py to execWithCallback().  The
execWithPulseProgress() function is now implemented using execWithCallback.
You can see how I use execWithCallback in the new dasd.py for displaying
progress as captured from the dasdfmt command.

I did introduce storage/dasd.py, which brings in a DASD class.  The whole
dasdfmt operation needs to occur before we initialize the storage code in
anaconda.  So like we have the iscsi and ZFCP classes and call startup()
on those before initializing the storage code, that's what the DASD class
is meant to do.  It also leaves a convenient place for future DASD patches
that need to deal with system-wide configuration files such as a possible
/etc/dasd.conf or something like that.

Lastly, compared to RHEL-5, the dasdfmt question is asked once.  We used
to present a Yes/Skip dialog for each unformatted DASD, which is simply
unusable.  I now build a list of all unformatted DASDs and tell the user
how many were found and ask if you want them formatted.  I did kick around
the idea of presenting a checklist where users could uncheck DASDs they
did not want anaconda to format and use, but really, we give users plenty
of ways to skip or mask out devices with all of the linuxrc.s390 stuff, I
felt doing it here was pointless.  By the time anaconda runs, we should
be able to use all of the DASDs we see.

The next patch set will be for the parallel formatting with dasdfmt.

Progress bar code in this set has been tested on both the iw and textw
code.  Works fine on both.

Both patches look good, I esp. like how this barely touches the main code
and has the bulk of the code in its own dasd.py. I do wonder about the
automatic formatting if there is no interface, this seems a bit of a
destructive default ?

Maybe we should offer a kickstart command to control this (and maybe
default to not formatting when that command is not present) ?

This kickstart can be added in a separate patch (if deemed desirable).

At first, the 'clearpart --initlabel' option sounded good, but I think there
should be a way for kickstart users to say "only run dasdfmt on disks without
a CDL format, leave formatted disks alone".  As stated in another reply, I'll
leave the kickstart stuff for another patch after I can talk to clumens.

- -- David Cantrell <dcantrell redhat com>
Red Hat / Honolulu, HI

Version: GnuPG v1.4.9 (GNU/Linux)


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