03.12.2018 18:12, Max Reitz wrote: > From: Vladimir Sementsov-Ogievskiy <[email protected]> > > This test is broken without previous commit fixing dead-lock in mirror. > > Signed-off-by: Vladimir Sementsov-Ogievskiy <[email protected]> > Signed-off-by: Max Reitz <[email protected]> > --- > Based-on: [email protected] > > This is a v3 to the second patch of "[PATCH v2 0/2] mirror dead-lock" by > Vladimir. Instead of mirroring from a fully allocated image, this > version mirrors from a preallocation=metadata image, and it mirrors not > to the real disk, but to a throttle/null-co combination. > > The advantages are that it is an indicator of the bug independently of > disk speed (Vladimir's version always worked on tmpfs for me), and that > it allocates less than 1 MB of space in contrast to 2 GB. > > The disadvantage is that it is not a real-life use case. We may want to > add Vladimir's test, too, probably in a special group that is not > executed by default. > > Also, this version fixes compatibility with python3 (by using > iotests.log()), and it kills the VM even on failure (by catching and > reraising the exception, after the VM has been shut down). > > Finally, the test can still be in the quick group, because it turns out > that bps-total=size is enough to make it fail before the fix (so it runs > in 1 s if it passes).
works for me, thank you! -- Best regards, Vladimir
