Merge ~canonical-kernel-team/+git/autotest-client-tests:phlin/xfstests-loopdev into ~canonical-kernel-team/+git/autotest-client-tests:master
Status: | Merged |
---|---|
Merged at revision: | 89adabb73aa7d5aebad6358d4da54545dacbe686 |
Proposed branch: | ~canonical-kernel-team/+git/autotest-client-tests:phlin/xfstests-loopdev |
Merge into: | ~canonical-kernel-team/+git/autotest-client-tests:master |
Diff against target: |
239 lines (+42/-24) 6 files modified
ubuntu_xfstests_btrfs/control.ubuntu (+6/-6) ubuntu_xfstests_btrfs/create-test-partitions (+8/-2) ubuntu_xfstests_ext4/control.ubuntu (+6/-6) ubuntu_xfstests_ext4/create-test-partitions (+8/-2) ubuntu_xfstests_xfs/control.ubuntu (+6/-6) ubuntu_xfstests_xfs/create-test-partitions (+8/-2) |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Francis Ginther | Approve | ||
Review via email:
|
Commit message
Use loop device for the test, courtesy to Francis who add this code
on AWS cloud first. This patch was ported from there and rebased
on the xfstest build fix (LP: #2008049), and use /dev/loop0 for
finding which free loop device to use.
By doing so we can avoid destroying data in /dev/sdb accidentally.
Also it can reduce the hardware limit of this test, it will no
longer require a scratch drive to run but just 100G free disk space.
Description of the change
The required 100G disk space might be debatable here. But I didn't try out to see how much disk space we need.
The next thing we need to do is:
* Add some limitation to this test, so that it won't be tested with smaller instances.
* Improve the test granularity to make it easier to read / debug with
* (Optional) Cherry-pick the test cases like what we did for ubuntu_
According to the XFSTests doc, the biggest disk exercised is 20GB ( This is only for BIGTEST and BIGSCRATCH tests & bigalloc_4k) /github. com/tytso/ xfstests- bld/blob/ master/ Documentation/ what-is- xfstests. md
https:/
I don't see a problem with lowering 100GB -> 20GB unless there is some urgent need for it.