Node install fails when sda has pre-existing md raid arrays

[Michael Jeanson]

With maas-region-controller 1.2+bzr1351+dfsg-0ubuntu1~precise1 from the stable ppa, node install will fail if the install drive has pre-existing md raid arrays. I believe the following should be added to /usr/share/maas/preseeds/preseed_master :

d-i partman-md/device_remove_md boolean true

[Andres Rodriguez (andreserl)]

Hi Michael,

While I do not have experience with md raid arrays, is this something we are looking on doing (destroying the array)? Or is this something that has to be done every time a new installation is done?

Cheers.

[Scott Moser (smoser)]

This seems logical to me.
The general philosophy is that every acquire -> release cycle is stand-alone. having one affect the other is not intended.

[Michael Jeanson (mjeanson)]

Hi Andres,

Long time no see,

I think the installation process shouldn't be altered by the content of the installation target disk, currently if the installer detects an md raid partition on the drive it will ask if it should continue or not and since that answer ins't preseeded it will basically hang.

In my case that only happens on the first install with mass since I use test servers that had md arrays from a previous project.

Maas already wipes the drive without any question whatever it's content, it should not be different for software raid arrays.

[Julian Edwards (julian-edwards)]

Hi there - if you propose your branch for merging I'll land it in trunk and 1.2 for you, so it should hit all releases in the imminent SRU.