In <20090429141142.ga19...@piper.oerlikon.madduck.net>, martin f krafft wrote: >also sprach Boyd Stephen Smith Jr. <b...@iguanasuicide.net> [2009.04.29.1557 +0200]: >> >One should thus fix LVM to be a bit more careful... >> >> LVM allows you to strictly limit what devices it scans for PV headers. > >That's not enough; LVM knows that md exists, and LVM-on-md is about >99.99998% of the sane use-cases, so LVM should require a --force or >even --yes-i-know-what-i-am-doing-so-let-the-monkeys-rain flag >before it uses a physical volume that has an md superblock.
I'm down with LVM running something like: mdadm --has-superblock /dev/block/device for devices that have a PV header and refusing to automatically treat them as PVs if it returns success, as long as it doesn't affect md-on-LVM. I haven't used it recently, but I generally do my luks/aesloop on top of LVM. The detection needs to be done by "md" itself, though, so that LVM doesn't have to have separately-maintained detection code that will eventually become out-of-date and leave us in the same situation. -- Boyd Stephen Smith Jr. ,= ,-_-. =. b...@iguanasuicide.net ((_/)o o(\_)) ICQ: 514984 YM/AIM: DaTwinkDaddy `-'(. .)`-' http://iguanasuicide.net/ \_/
signature.asc
Description: This is a digitally signed message part.
