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

Re: [libvirt] [PATCH] Do not try to cancel non-existent migration on source

On 08/25/2011 01:03 PM, Jiri Denemark wrote:
On Tue, Aug 16, 2011 at 21:39:41 +0200, Jiri Denemark wrote:
On Tue, Aug 16, 2011 at 08:33:04 -0600, Eric Blake wrote:
On 08/16/2011 04:44 AM, Jiri Denemark wrote:
If migration failed on source daemon, the migration is automatically
canceled by the daemon itself. Thus we don't need to call
virDomainMigrateConfirm3(cancelled=1). Calling it doesn't cause any harm
but the resulting error message printed in logs may confuse people.
   src/libvirt.c |   41 +++++++++++++++++++++++++----------------
   1 files changed, 25 insertions(+), 16 deletions(-)

Shouldn't qemu_migration.c get the same fix, since peer2peer and
tunneled migration basically re-implement the same migration driver

It's reimplementing the same thing but in a different context. The code in
libvirt.c runs on a client, while qemu_migration.c runs within the qemu driver
in libvirtd. The virDomainMigratePrform3 API called from libvirt.c cleans up
after itself; it resumes a domain and finishes the migration job in case of
error. On the other hand, everything in doPeer2PeerMigrate3 is run within a
single public API and the function called to do the Perform phase is an
internal function which doesn't clean up after itself in the same way as the
public API has to. Hence, we need to call qemuMigrationConfirm to actually
resume the domain even if the Perform phase failed.

Ping. Did I manage to convince you this patch is complete? :-)

Yes. ACK, although since I just pushed Alex' patch, you might have a merge conflict to fix when you rebase.

Eric Blake   eblake redhat com    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

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