On Thu, Feb 22, 2018 at 07:19:39PM +0100, Tomáš Golembiovský wrote:
The previous patches seem OK. On a quick glance this one also seems
a
good start.
On Thu, 22 Feb 2018 13:57:25 +0000
"Richard W.M. Jones" <rjones(a)redhat.com> wrote:
> PROBLEMS:
> - Spools to a temporary disk
> - Need to specify direct/indirect upload via flag
> - Location of ca.pem
We surely need a new argument for that.
When running on VDSM host the certificate is in
'/etc/pki/vdsm/certs/cacert.pem'. Maybe we can use it as default?
OK.
> - Target cluster
> - Delete disks on failure, or rename disks on success?
Definitely delete disks on failure. Unless we want to make it
administrators task. If you want to mark disks created by upload maybe
you can put something like 'uploaded by virt-v2v' in a comment? That may
help avoiding any renames... just an idea.
So I guess the first question is _can_ we rename disks?
It's not always possible for virt-v2v to delete disks on every
conceivable failure -- eg. if we're linked into a C library and that
segfaults, there is no recovery.
If we could rename disks then we could give them temporary names
during the upload part, and rename then right at the very end. This
would greatly reduce the window where we could leave an unused disk
around. It would mean that something would need to periodically scan
RHV to look for the temporary disks and delete them (perhaps a single
job after the whole conversion has finished?)
Rich.
--
Richard Jones, Virtualization Group, Red Hat
http://people.redhat.com/~rjones
Read my programming and virtualization blog:
http://rwmj.wordpress.com
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine. Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/