Re: Non-deterministic ordering of raid partition creation...

["James Olin Oden" <james oden gmail com>]

On 12/12/06, Oden, James <James Oden tekelec com> wrote:
> So I have a Kickstart file that is generating about 12 raid partitions
> across 2 drives.  What I'm finding is that it does not create them in
> the order I specify.  Is this a known problem?
>
> Other than that, I'm about to go in and try to figure out why this is
> happening, any ideas what code would be best to start looking at?
Here is the section that requests all the raid partitions:

  zerombr yes
  clearpart --drives=hda,hdb --all --initlabel
  part raid.11  --size=256   --ondrive=hda --asprimary
  part raid.12  --size=512   --ondrive=hda --asprimary
  part raid.13  --size=2000  --ondrive=hda --asprimary
  part raid.15  --size=4096  --ondrive=hda
  part raid.16  --size=1024  --ondrive=hda
  part raid.17  --size=512   --ondrive=hda
  part raid.18  --size=4096  --ondrive=hda
  part raid.19  --size=512   --ondrive=hda
  part raid.110 --size=4096  --ondrive=hda
  part raid.111 --size=15500 --ondrive=hda
  part raid.112 --size=7119    --ondrive=hda
  part raid.21  --size=256   --ondrive=hdb --asprimary
  part raid.22  --size=512   --ondrive=hdb --asprimary
  part raid.23  --size=2000  --ondrive=hdb --asprimary
  part raid.25  --size=4096  --ondrive=hdb
  part raid.26  --size=1024  --ondrive=hdb
  part raid.27  --size=512   --ondrive=hdb
  part raid.28  --size=4096  --ondrive=hdb
  part raid.29  --size=512   --ondrive=hdb
  part raid.210 --size=4096  --ondrive=hdb
  part raid.211 --size=15500 --ondrive=hdb
  part raid.212 --size=7119    --ondrive=hdb
  raid /boot                  --fstype=ext3 --level=1 --device=1
raid.11  raid.21
  raid /                      --fstype=ext3 --level=1 --device=2
raid.12  raid.22
  raid swap                   --fstype=swap --level=1 --device=3
raid.13  raid.23
  raid /usr                   --fstype=ext3 --level=1 --device=5
raid.15  raid.25
  raid /tmp                   --fstype=ext3 --level=1 --device=6
raid.16  raid.26
  raid /var                   --fstype=ext3 --level=1 --device=7
raid.17  raid.27
  raid /var/TKLC              --fstype=ext3 --level=1 --device=8
raid.18  raid.28
  raid /var/TKLC/recovery     --fstype=ext3 --level=1 --device=9
raid.19  raid.29
  raid /var/TKLC/recovery/usr --fstype=ext3 --level=1 --device=10
raid.110 raid.210
  raid /var/TKLC/swap         --fstype=ext3 --level=1 --device=11
raid.111 raid.211
  raid pv.01 --device=md12 --level=1 raid.112 raid.212
  volgroup vgroot pv.01

And the partitioning we end up with looks like:

                     REAL         EXPECTED
      raid.11     256M          256M
      raid.12     2000M        512M
      raid.13     512M          2000M
      raid.15     15500M      4096M
      raid.16     7119M        1024M
      raid.17     4096M        512M
      raid.18     4096M        4096M
      raid.19     4096M        512M
      raid.110   1024M        4096M
      raid.111   512M          15500M
      raid.112   512M           7119M


Which basically does not resemble the order of the raid partitions at
all?  What am I missing?  Is this a bug (it feels like one to me)?
Bottom line how do I tell to anaconda generate raid partitions in some
deterministic order or what part of the anaconda code do I need to be
looking at in order to fix this?

FYI, I am looking in partition.py and partRequest.py, but I haven't
quite unraveled the code path that goes from parse, to generating
little partition objects to creating them yet.

Thanks...james