Main inclusion report.

[Luke Yelavich]

Binary package hint: dmraid

Wish to include in main to allow for users with fakeraid configurations, such as is used with windows, to be able to install Linux onto the same fakeraid setup.

https://wiki.ubuntu.com/MainInclusionReportDmraid

[Matthias Klose (doko)]

$ find -name '*.orig'
./lib/format/ddf/ddf1.c.orig

- lib/format/ataraid/jm.c (name): len is shortened to JM_NAME_LEN, but then the full `name' is copied into the buffer.

Packaging looks ok, please check the above problem.

Could problems arise with existing installations now that we support dmraid?

[Luke Yelavich (themuso)]

The .orig file is in the upstream tarball.

Re the jm.c weirdness, I've actually found a patch from upstream that cleans that code up a lot. I've applied this patch, and have successfully built and uploaded it.

[Luke Yelavich (themuso)]

As for possible problems with dmraid installed on machines, I would think it would only get installed if the user needs it for a setup, similar to the way lvm2 and mdadm are only installed when they are needed. Added to that, dmraid only sets up raid sets if it can find valid metadata on hard disks, written bya particular fakeraid bios. Even if it finds such metadata, since we now use UUIDs for root devices and other partitions, I don't think dmraid will cause a problem with causing partitions to be mounted from an incorrect device.

[Martin Pitt (pitti)]

Luke Yelavich [2008-02-15 0:25 -0000]:
> As for possible problems with dmraid installed on machines, I would
> think it would only get installed if the user needs it for a setup,
> similar to the way lvm2 and mdadm are only installed when they are
> needed.

That's not quite true, the packages are installed by default
unconditionally. The same will happen with dmraid.

[Tormod Volden (tormodvolden)]

> That's not quite true, the packages are installed by default
> unconditionally. The same will happen with dmraid.

When I installed Hardy Alpha 4 from scratch (but with existing LVM volumes on the disk) I had to install lvm2 manually. This was the Desktop installer, I don't know if you're talking about the Alternate or both.

[Martin Pitt (pitti)]

Luke did some analysis on the packages and found that it is still too immature to be supportable. For example, there is no notification at all when a component of the RAID gets defective, and it is not possible to rebuild a degraded array with the current packages.

Rejecting for hardy. This should be reconsidered for Hardy+1 once the packages get more mature and usable.

[Luke Yelavich (themuso)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Re-opening, per the DmRaid Support spec: https://wiki.ubuntu.com/DmraidSupport

 affects ubuntu/dmraid
 status new
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFIkTSMjVefwtBjIM4RAgGjAJ0QwYWoVDFSRAqLsYDgrTl+SRzSFACgkhL2
XfSClub/mhmWzcqpRV1l/ec=
=oR2N
-----END PGP SIGNATURE-----

[Colin Watson (cjwatson)]

I'm accepting this and have promoted it to main. Luke is well on top of this and has addressed the bulk of the issues raised to date, including work on improving handling degraded arrays. I'm satisfied with the progress made thus far and I think it's worth unleashing it on Ubuntu users for further testing in Intrepid's stabilisation period.

(Obviously I have no objection to full-time MIR reviewers re-reviewing this, but at the time of writing both Martin and Matthias were on holiday and I wanted to unblock this for feature freeze.)