[Qemu-devel] [PATCH] migration: yet more possible state transitions

[Qemu-devel] [PATCH] migration: yet more possible state transitions

[Juan Quintela]

On destination, we move from INMIGRATE to FINISH_MIGRATE.  Add that to
the list of allowed states.

Signed-off-by: Juan Quintela <quintela@redhat.com>
---
 vl.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/vl.c b/vl.c
index 0adbbd6..45eb9ea 100644
--- a/vl.c
+++ b/vl.c
@@ -582,6 +582,7 @@ static const RunStateTransition runstate_transitions_def[] = {
     { RUN_STATE_INMIGRATE, RUN_STATE_SUSPENDED },
     { RUN_STATE_INMIGRATE, RUN_STATE_WATCHDOG },
     { RUN_STATE_INMIGRATE, RUN_STATE_GUEST_PANICKED },
+    { RUN_STATE_INMIGRATE, RUN_STATE_FINISH_MIGRATE },

     { RUN_STATE_INTERNAL_ERROR, RUN_STATE_PAUSED },
     { RUN_STATE_INTERNAL_ERROR, RUN_STATE_FINISH_MIGRATE },
-- 
2.4.3

Re: [Qemu-devel] [PATCH] migration: yet more possible state transitions

[Juan Quintela]

Juan Quintela <quintela@redhat.com> wrote:
> On destination, we move from INMIGRATE to FINISH_MIGRATE.  Add that to
> the list of allowed states.
>
> Signed-off-by: Juan Quintela <quintela@redhat.com>

Self-NAK.

Somewhere I was able to test old + new qemu given this weird behaviour.
behaviour makes no sense, and I am not able to reproduce anymore.

Sorry for the noise.

> ---
>  vl.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/vl.c b/vl.c
> index 0adbbd6..45eb9ea 100644
> --- a/vl.c
> +++ b/vl.c
> @@ -582,6 +582,7 @@ static const RunStateTransition runstate_transitions_def[] = {
>      { RUN_STATE_INMIGRATE, RUN_STATE_SUSPENDED },
>      { RUN_STATE_INMIGRATE, RUN_STATE_WATCHDOG },
>      { RUN_STATE_INMIGRATE, RUN_STATE_GUEST_PANICKED },
> +    { RUN_STATE_INMIGRATE, RUN_STATE_FINISH_MIGRATE },
>
>      { RUN_STATE_INTERNAL_ERROR, RUN_STATE_PAUSED },
>      { RUN_STATE_INTERNAL_ERROR, RUN_STATE_FINISH_MIGRATE },

Re: [Qemu-devel] [PATCH] migration: yet more possible state transitions

[Paolo Bonzini]

On 28/07/2015 17:19, Juan Quintela wrote:
>> > On destination, we move from INMIGRATE to FINISH_MIGRATE.  Add that to
>> > the list of allowed states.
>> >
>> > Signed-off-by: Juan Quintela <quintela@redhat.com>
> Self-NAK.
> 
> Somewhere I was able to test old + new qemu given this weird behaviour.
> behaviour makes no sense, and I am not able to reproduce anymore.

I think it makes sense.  Did you migrate twice on the source?  Then the
state was FINISH_MIGRATE at the beginning of the second migration.

Paolo

Re: [Qemu-devel] [PATCH] migration: yet more possible state transitions

[Juan Quintela]

Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 28/07/2015 17:19, Juan Quintela wrote:
>>> > On destination, we move from INMIGRATE to FINISH_MIGRATE.  Add that to
>>> > the list of allowed states.
>>> >
>>> > Signed-off-by: Juan Quintela <quintela@redhat.com>
>> Self-NAK.
>> 
>> Somewhere I was able to test old + new qemu given this weird behaviour.
>> behaviour makes no sense, and I am not able to reproduce anymore.
>
> I think it makes sense.  Did you migrate twice on the source?  Then the
> state was FINISH_MIGRATE at the beginning of the second migration.

It could be, I would try again with normal trees (I was using and old
compiled binary in one of the sides when errors happened).

Later, Juan.