Bug #1778
closedError after installing an iso-image via qemu / rbd-image
0%
Description
Hi *,
we are currently running:
ceph version 0.38 (commit:b600ec2ac7c0f2e508720f8e8bb87c3db15509b9) from deb-pkg: Version: 0.38-1~bpo60+1
on Linux-kernel 3.1.2 with neccessary modules loaded.
Further setup:
- QEMU emulator version 0.15.94, Copyright (c) 2003-2008 Fabrice Bellard
With some running VM's we have evaluated the snapshot-capability, which worked like
a charme managed via the qemu monitoring-port ( savevm, loadvm...). So far, so very
good.
And now to the paragraph containing the long expected "But":
But, as of a fresh os-install into a rbd-image from for example an
Ubuntu-10.04.1-server-amd64.iso
everything went fine up to a complete stop/start.
No disk was found. After some re-installs we have a very early failure even
after an "rbd-map" of the image:
"rbd0: unknown partition table"
which went away after not starting the VM with the "snapshot=on" parameter for the
corresponding drive and reinstalling the same iso.
So, our conclusion leads to s/t concerning the snapshot-functionality reporting some
false geometry-values to fdisk n grub n friends, or wrong start-cyls which should be
after snapshot-meta-data but are not?! ( just guessing)
Well, as we have no other clues right now, please sched some light in ;)
We are of course willing to assist with any debugging.
Regards and thnx in@vance,
Oliver.
Files