Currently, you cannot cancel a mirror job without the @force flag set.
This is intentional once source and target are in sync, but probably not
so much before that happens.  The main reason for me thinking this is
because it is an undocumented change in 2.12.0 in respect to 2.11.

(b76e4458b1eb3c32e9824fe6aa51f67d2b251748 only notes a behavior change
 after READY, but not before.)


This series depends on Stefan’s patch “block/mirror: honor ratelimit
again” (which is in Jeff’s queue).

Based-on: <20180424123527.19168-1-stefanha@redhat.com>


Max Reitz (2):
  block/mirror: Make cancel always cancel pre-READY
  iotests: Add test for cancelling a mirror job

 block/mirror.c             |   4 +-
 tests/qemu-iotests/218     | 138 +++++++++++++++++++++++++++++++++++++++++++++
 tests/qemu-iotests/218.out |  30 ++++++++++
 tests/qemu-iotests/group   |   1 +
 4 files changed, 172 insertions(+), 1 deletion(-)
 create mode 100755 tests/qemu-iotests/218
 create mode 100644 tests/qemu-iotests/218.out

-- 
2.14.3

On 05/01/2018 05:05 PM, Max Reitz wrote:
> Currently, you cannot cancel a mirror job without the @force flag set.
> This is intentional once source and target are in sync, but probably not
> so much before that happens.  The main reason for me thinking this is
> because it is an undocumented change in 2.12.0 in respect to 2.11.

Yeah, that definitely feels like a regression worth fixing.

> 
> (b76e4458b1eb3c32e9824fe6aa51f67d2b251748 only notes a behavior change
>   after READY, but not before.)
> 
> 
> This series depends on Stefan’s patch “block/mirror: honor ratelimit
> again” (which is in Jeff’s queue).
> 
> Based-on: <20180424123527.19168-1-stefanha@redhat.com>
> 


-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

On Wed, May 02, 2018 at 12:05:07AM +0200, Max Reitz wrote:
> Currently, you cannot cancel a mirror job without the @force flag set.
> This is intentional once source and target are in sync, but probably not
> so much before that happens.  The main reason for me thinking this is
> because it is an undocumented change in 2.12.0 in respect to 2.11.
> 
> (b76e4458b1eb3c32e9824fe6aa51f67d2b251748 only notes a behavior change
>  after READY, but not before.)
> 
> 
> This series depends on Stefan’s patch “block/mirror: honor ratelimit
> again” (which is in Jeff’s queue).
> 
> Based-on: <20180424123527.19168-1-stefanha@redhat.com>
> 
> 
> Max Reitz (2):
>   block/mirror: Make cancel always cancel pre-READY
>   iotests: Add test for cancelling a mirror job
> 
>  block/mirror.c             |   4 +-
>  tests/qemu-iotests/218     | 138 +++++++++++++++++++++++++++++++++++++++++++++
>  tests/qemu-iotests/218.out |  30 ++++++++++
>  tests/qemu-iotests/group   |   1 +
>  4 files changed, 172 insertions(+), 1 deletion(-)
>  create mode 100755 tests/qemu-iotests/218
>  create mode 100644 tests/qemu-iotests/218.out
> 
> -- 
> 2.14.3
> 

Thanks,

Applied to my block branch:

git://github.com/codyprime/qemu-kvm-jtc block

-Jeff

On 05/01/2018 06:05 PM, Max Reitz wrote:
> Currently, you cannot cancel a mirror job without the @force flag set.
> This is intentional once source and target are in sync, but probably not
> so much before that happens.  The main reason for me thinking this is
> because it is an undocumented change in 2.12.0 in respect to 2.11.
> 
> (b76e4458b1eb3c32e9824fe6aa51f67d2b251748 only notes a behavior change
>  after READY, but not before.)
> 
> 
> This series depends on Stefan’s patch “block/mirror: honor ratelimit
> again” (which is in Jeff’s queue).
> 
> Based-on: <20180424123527.19168-1-stefanha@redhat.com>
> 
> 
> Max Reitz (2):
>   block/mirror: Make cancel always cancel pre-READY
>   iotests: Add test for cancelling a mirror job
> 
>  block/mirror.c             |   4 +-
>  tests/qemu-iotests/218     | 138 +++++++++++++++++++++++++++++++++++++++++++++
>  tests/qemu-iotests/218.out |  30 ++++++++++
>  tests/qemu-iotests/group   |   1 +
>  4 files changed, 172 insertions(+), 1 deletion(-)
>  create mode 100755 tests/qemu-iotests/218
>  create mode 100644 tests/qemu-iotests/218.out
> 

Way too late to the party, but all the same, since I was CC'd and I
already looked into it:

Reviewed-by: John Snow <jsnow@redhat.com>