Re: [PATCH 0/7] qemu: tpm: Add support for migration across shared storage

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, Aug 22, 2022 at 02:52:26PM -0400, Stefan Berger wrote:
> On 8/22/22 12:35, Daniel P. Berrangé wrote:
> > On Mon, Aug 22, 2022 at 08:05:47AM -0400, Stefan Berger wrote:
> > > This series of patches adds support for migrating vTPMs across hosts whose
> > > storage has been set up to share the directory structure holding the state
> > > of the TPM (swtpm). The domain XML is extended with a shared_storage
> > > attribute that must be set to 'yes' when shared storage is used. It
> > 
> > We don't require any 'shared_storage' attribute for disk images - we just
> > aim to "do the right thing" automatically. If we want to support shared
> > storage for TPM, then IMHO it should likewise be made transparent.
> > 
> > What's the thinking behind putting the TPM on shared storage though ?
> 
> It's drive by swtpm user(s):
> 
> https://github.com/stefanberger/swtpm/pull/732
> 
> The driving force is having the state available on the destination to
> restart a VM there if the original host failed. Allegedly all hosts in their
> setup would share the necessary storage to be able to do that with TPM state
> but then presumably also with the disk image(s).

Ok, so use case is resilience rather than specifically live
migration.


> > The state is tiny so you're not going to notice it being transferred
> 
> Tiny is relative to disk sizes. It can become ~260kb or so, depending on how
> much is stored in the NVRAM areas, but yes, it's comparably small.

I meant 'tiny' in the sense that the time required to live migration
it is not measureably significant.  Compared with say migrating disk
storage, which could add hours to a live migration if it wasn't on
shared storage.

> > in the regular migration stream, so there's no performance benefit
> > here.  Meanwhile it hurts security because we can't do SELinux isolation
> > on many NFS install, and the complexity of the dance around locking makes
> > the migration process more fragile. The pain we've had dealing with NFS
> > makes it really unappealing
> 
> 
> Others would want to use CephFS for it I heard and I am not sure what the
> pain points with this shared storage tech are.
> 
> > 
> > > influences the management of the directory structure holding the TPM state,
> > > which for example is only to be removed when a domain is undefined (virsh
> > > undefine) and not when a VM is removed on the migration source host.
> > > Further, when shared storage is used security labeling on the destination
> > > side is skipped assuming that the labeling was already done on the source
> > > side.
> > > 
> > > I have tested this with an NFS setup where I had to turn SELinux off on
> > > the hosts since the SELinux MLS range labeling is not supported.
> > 
> > Have you tested in systems where 'root_squash' is enabled on the NFS
> > server, such that libvirt can't create the files as 'root' and then
> > chown them to 'qemu' later ? That's been a major source of pain related
> > to NFS for disks and snapshots historically.
> 
> Yes, that doesn't seem to work. I had used it with no_root_squash before.
> 
> # virsh start testVM
> error: Failed to start domain 'testVM'
> error: internal error: Could not create directory
> /var/lib/libvirt/swtpm/ecc221c4-6bb9-423f-ac31-72244fdbb1a1/tpm2 as 59:59
> 
>    Stefan
> 

With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|




[Index of Archives]     [Virt Tools]     [Libvirt Users]     [Lib OS Info]     [Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [KDE Users]     [Fedora Tools]

  Powered by Linux