Hi all!
Recently we faced the following task:
Customer comes and say: incremental backup images are too fat. Does you
incremental backup works correct?
What to answer? We should check something. At least check that
incremental images doesn't store same data twice. And we don't have a
tool for it. I just wrote a simple python script to compare raw files
cluster-by-cluster. Then we've mounted the qcow2 images with help of
qemu-nbd, the resulting /dev/nbd* were compared and we proved that
incremental backups don't store same data.
Good idea. I love diagnostic tools!
But that leads to idea that some kind of that script would be good to
have at hand.
So, here is a new option for qemu-img compare, that is a lot more
powerful and effective than original script, and allows to compare and
calculate statistics, i.e. how many clusters differs, how many
clusters changed from unallocated to data, and so on.
For examples of output look at the test in patch 04.
Vladimir Sementsov-Ogievskiy (4):
qemu-img: implement compare --stat
qemu-img: make --block-size optional for compare --stat
qemu-img: add --shallow option for qemu-img compare --stat
iotests: add qemu-img-compare-stat test
docs/tools/qemu-img.rst | 29 +-
qemu-img.c | 275 +++++++++++++++++-
qemu-img-cmds.hx | 4 +-
.../qemu-iotests/tests/qemu-img-compare-stat | 88 ++++++
And new tests! :-)
.../tests/qemu-img-compare-stat.out | 106 +++++++
5 files changed, 484 insertions(+), 18 deletions(-)
create mode 100755 tests/qemu-iotests/tests/qemu-img-compare-stat
create mode 100644 tests/qemu-iotests/tests/qemu-img-compare-stat.out