[KVM-AUTOTEST][PATCH] Win2003-64.steps: fix a barrier at step 17 that sometimes fails due to popups.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com Signed-off-by: Michael Goldish mgold...@redhat.com diff --git a/client/tests/kvm_runtest_2/steps/Win2003-64.steps b/client/tests/kvm_runtest_2/steps/Win2003-64.steps index 2ce794f..9a4d3e7 100644 --- a/client/tests/kvm_runtest_2/steps/Win2003-64.steps

[KVM-AUTOTEST][PATCH] kvm_guest_wizard.py, stepmaker.py: do not explicitly print failure messages.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com This is now taken care of by kvm_runtest_2.py. Signed-off-by: Michael Goldish mgold...@redhat.com diff --git a/client/tests/kvm_runtest_2/kvm_guest_wizard.py b/client/tests/kvm_runtest_2/kvm_guest_wizard.py index 2f08c78..afe6b02 100644 ---

[KVM-AUTOTEST][PATCH] kvm_guest_wizard.py: barrier_2: sleep at the end of each iteration.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com Sleep at the end of the iteration, after checking the md5sum for the first time. This enables us to pass the barrier very quickly in some cases. (Currently we sleep first and check the md5sum later.) Signed-off-by: Michael Goldish mgold...@redhat.com

[KVM-AUTOTEST][PATCH] RHEL-3.9 step files: fix the first barriers, which are a little risky.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com In these step files, the regions picked for the barriers are updated before the regions that actually indicate readiness for keyboard input (the boot: prompt). Therefore, when the step files run, keyboard input (Enter key) is given before the VM is ready,

[KVM-AUTOTEST][PATCH] make_html_report.py: don't use head -n 1 when getting the KVM version.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com Use python's splitlines() instead of head -n 1 because the latter causes broken pipe messages in some configurations. Signed-off-by: Michael Goldish mgold...@redhat.com diff --git a/client/tests/kvm_runtest_2/make_html_report.py

[KVM-AUTOTEST][PATCH] Fedora-8-i386.steps: fix a barrier at step 26 that sometimes fails.

2009-04-19 Thread Uri Lublin
From: Alexey Eromenko aerom...@redhat.com Signed-off-by: Alexey Eromenko aerom...@redhat.com Signed-off-by: Michael Goldish mgold...@redhat.com diff --git a/client/tests/kvm_runtest_2/steps/Fedora-8-i386.steps b/client/tests/kvm_runtest_2/steps/Fedora-8-i386.steps index b23878a..732d705 100644

[KVM-AUTOTEST][PATCH] WinXP-32.steps: add an optional barrier to deal with closed start menu.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com Usually the barrier at step 24 should find an open start menu. Sometimes the start menu is closed, so an additional step is needed to open it in that case. Signed-off-by: Michael Goldish mgold...@redhat.com diff --git

[KVM-AUTOTEST][PATCH] kvm_tests.cfg.sample: fix typo in RHEL.4.7.x86_64 ISO filename.

2009-04-19 Thread Uri Lublin
From: Alexey Eromenko aerom...@redhat.com Signed-off-by: Alexey Eromenko aerom...@redhat.com Signed-off-by: Michael Goldish mgold...@redhat.com Signed-off-by: Uri Lublin u...@redhat.com diff --git a/client/tests/kvm_runtest_2/kvm_tests.cfg.sample

[KVM-AUTOTEST][PATCH] kvm_tests.cfg.sample: Windows: change migration_test_command to ver vol.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com The current migration_test_command is help, which for some reason returns a non-zero %errorlevel%. This makes get_command_status_output report the output as if an error has occurred. ver vol returns 0, and is also probably more informative.

[KVM-AUTOTEST][PATCH] kvm_tests.cfg.sample: remove -snapshot from boot and reboot tests.

2009-04-19 Thread Uri Lublin
From: Michael Goldish mgold...@redhat.com It shouldn't be too risky to run these tests without -snapshot. It will also save some time because the following tests, which run without -snapshot, won't have to restart the VM. That's true since following tests do not use -snapshot, so the framework

[KVM-AUTOTEST][PATCH] kvm_install: make buildir (where kvm is installed) next to srcdir

2009-04-19 Thread Uri Lublin
From: Uri Lublin u...@redhat.com If the user specify a directory for the source (srcdir) it makes sense that buildir would reside in the same directory. This is useful for server dispatched client tests, as without this patch one would have to re-install kvm for every job. Note: this works for