libvirt-tests: Fix some emulatorpin errors

libvirt: Add virsh net-info command test

run: Add --no-cleanup option

For debugging purposes, not cleaning up env files,
test tmp files and aexpect tmp files might be desirable.
So add this option to our test runner.

CC: Lukas Doktor <ldoktor@redhat.com>
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

JeOS 19 ks: Fill up disk image so empty space can be removed

Add a last step post install that will make the JeOS
image to fill all 10GB of disk, so that we can easily
create an image with no empty space by using qemu-img

qemu-img convert -O qcow2 jeos-19-x86_64.qcow2 jeos-19-x86_64-output.qcow2
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

JeOS 19: Add qemu-guest-agent

It was argued that the agent is closely related to
virtualization testing, and it is small enough to
justify inclusion.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

virsh_snapshot_edit: Resolve test errors

virsh_snapshot_create_as: Adjustments from review

virsh.schedinfo_qemu_posix: update the way we get the cgroup path

Make the test runner more stable with 'safe' options

staging.utils_cgroup: change the regexp for finding correct cgroup path

Add test for virsh find-storage-pool-sources-as and bug fix

Mq change qnum

On an attempt to eliminate sources of disk corruption
and errors, delete the dst file if the src file is
avaliable.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Use SIGKILL instead of SIGTERM (the default for pkill)
when terminating the background process. This way we're
less likely to run into a situation where killing the
tcpdump command times out.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Sometimes, post migration run, we can run into a condition
where tcpdump or the VM hangs at migration cleanup stage.
After analyzing test logs, including VM logs, there's nothing
that points out to any guest OS crash or similar condition.

With that in mind, it is admissible to be more lenient
when trying to kill the migration background command, and
if a timeout happens when trying to kill it, be safe and
try to shut down the vm gracefully.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

As the pre processing code already takes care of that
anyway.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

This class was logging a debug event every time it successfully
reused an aexpect session.  Since aexpect always loggs the command
being run anyway, this extra detail is mostly redundant.  If there's
a problem, an exception will be thrown anyway.
Signed-off-by: NChris Evich <cevich@redhat.com>

When using original sync of NetworkXML class to restore a backup
network, it is redefined and started, leaving autostart disabled.
This is not the default setting (all true for active/persistent/
autostart) for the network.

In this modification, we set the default behavior of sync to
set all state to true. With the help of an optional "state" dict
parameter, we can restore the network to any state you like, or
in practical, to the state we backed up.

To facilitate this backup & restore process, I also added a helper
function state_dict() to retrieve a active/persistent/autostart
dict for backing up current vm state.

The unittest for these network manipulators is given in a new file
instead of libvirt_xml_unittest.py to separate manipulator tests
from XML base test.

Furthermore, The net_state_dict function use an outdated way to get
the persistent property for the network, which is already changed
by a patch (commit 895913d) in libvirt-0.10.2. The persistent status
can be retrieved from net_list directly. So I add a condition to
ensure it can be backported for older libvirt while avoiding
redundant call of net_autostart when testing on new libvirt.

This backport can be removed if old libvirt no longer maintained.
Signed-off-by: NHao Liu <hliu@redhat.com>

virttest.env_process: Remove restart_vm option

The restart_vm option, that will restart a vm at the beginning
of postprocessing, is unnecessary, given that you could
simply shut down all the VMs at the end of each test,
forcing new vms to be started.

So get rid of this option, and replace --restart-vm
in ./run to use kill_vm = yes instead.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

When run after virsh.shutdown, the first virsh.start was failing with
because the domain was already started.  Adjusted the .cfg file to
ensure setting of the 'kill_vm_before_test' in the status_error_no
variant.

Also, changed the error returned from using 'stdout' to 'stderr' so that
"Domain is already active" would be displayed.

Signed-off-by: NYanbing Du <ydu@redhat.com>

  * Wrong using 'elif' instead of 'if', then it will never judge if this target using
    the emulated image name.
  * Delete the duplicated judgment when cleanup.
Signed-off-by: NYanbing Du <ydu@redhat.com>

PEP8 mass auto fix

Catching up with some problems found w.r.t. coding style
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

qemu.tests.virtio_console: Update test_hotplug_virtio_pci

Fix the problem with duplicated IDs caused by early device_del execution.
Monitor returns no output but internally waits for OS ACPI respone. This
message, in case kernel is not yet initialized, never occurs.

This version resends the device_del in the first round until it succeeds,
than it just waits until the device disappears.

Additionally the test was rewritten to use qemu_devices to handle the
hotplug/unplug and the test cfg was tweaked to be more intensive.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Proper identification in qemu_devices is object.get_qid().
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Test was failing because the 're.search()' wasn't stripping off the
newline character from the after_xml, thus not allowing the match check
to determine if the <description> changed. Changed the error processing
to list the 15 or so relevant lines from the snapshot dump xml that
could be different if the re.search() fails.

Remove the extraneous .strip() from the after_xml creation since it's
repeated in the snap_xml_compare() method when doing the comparison
with the pre_xml. Also changed the logic on error path to just print
some different lines rather than printing both whole pre/after xml
and forcing an eyeball comparison for why the test failed.

virt: Fix hanging tail_threads which are created by recreating env.

A review comment/suggestion from @liyangfnst after pull request #1174
was merged to master was to make the entire test use try/finally type
logic so as to only run the xml_recover from one place...

This change implements that concept.

Autotest regression test

Libguestfs: supplements of guestmount test.

Signed-off-by: NJiří Župka <jzupka@redhat.com>

virttest: Add disable for warning color

If using the github repo URL. If not, default to the script on the github
master branch.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Since the next branch does not exist anymore.
Signed-off-by: NCleber Rosa <crosa@redhat.com>