existing raid superblocks confuse/break the install
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
partman-md |
New
|
Undecided
|
Unassigned | ||
partman-md (Ubuntu) |
Confirmed
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: partman-md
Partman fails if you try to set up raid with existing raid superblocks on some devices. (i.e. on re-installs/
Partman may get confused as raid devices get set up after partman (re)creates a partition and it happens that there is an superblock (at the beginning or end). Udev/mdadm sets it (partially) up and devices are in use.
The common workaround suggested is to make sure to allways "mdadm --zero-superblock" raid partitions/devices prior to deleting them, or to to "just zero the entire disk between reinstallation attempts" (often heard after bad experiences).
But partman-md should probably really watch out for/expect to see md devices being set up, (with udev/mdadm even after booting) and take them into account. (list them, offer to --stop, re-assemble (use) them as a real non-auxiliary device (not md127 and up), --zero-superblock if needed, ...)
related:
Bug #22301
Bug #33117
Bug #569900
description: | updated |
Changed in partman-md (Ubuntu): | |
status: | New → Confirmed |