Add lvm support for qemu tests

virttest.qemu_vm: Log information about the parent PID and the monitor n...

When starting the qemu process, there is not much information about
the running parent PID (the autotest shell script that starts qemu) or
the monitor name that we are trying to connect (for example, hmp1).

So with this commit we're logging this information when qemu is started,
in order to provide information for debugging. Example:

When running  ./run -t qemu --tests=uptime --qemu_sandbox=on

From debug.log:

11:43:17 INFO | Created qemu process with parent PID 15592
11:43:17 INFO | Connecting to monitor 'hmp1'
11:44:17 WARNI| Could not find (qemu) prompt after connecting to monitor. Output so far: ''
11:44:18 ERROR| Could not connect to monitor 'hmp1'
11:44:18 DEBUG| Destroying VM virt-tests-vm1 (PID 15594)

Parent PID is 15592, now we are trying to find what is wrong...

ps -ef | grep 15592
rmoura   15592 15591  0 11:43 pts/3    00:00:00 /bin/bash -c source /tmp/autotestbMRLRL.sh
rmoura   15594 15592  0 11:43 pts/3    00:00:00 [qemu-system-x86] <defunct>

In this situation, it's clear that qemu-system-x86_64 goes bad.
Signed-off-by: NRuda Moura <rmoura@redhat.com>

Signed-off-by: NXu Tian <xutian@redhat.com>

Signed-off-by: NXu Tian <xutian@redhat.com>

Signed-off-by: NXu Tian <xutian@redhat.com>

Signed-off-by: NXu Tian <xutian@redhat.com>

register lvm object to env file when use lvm storage, and restore it at
postprocess to clean up new volumes when "force_remove_image = yes";

Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

In fact, previous fixes that make nfs to use the
staging library needed some tweaking to match
expectations.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Make the nfs module use the service API previous to the
change made on 69ac938d.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Turns out that change was made with little thought
with regards to the users of the code that already
existed.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

virttest.qemu_devices: Make IDE bus strict_mode only

virsh_snapshot_create_as: Properly format output_dump

When using IDE bus, only ide.1 bus is used by default. We have to
force the ide.0 in order to use it (similar to AHCI). This patch
makes all QBusUnitBuses act this way.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Recent changes to dumpxml missed the snapshot_dumpxml reference
when changing the returned data as a CmdResult.  Format the output_dump
with the .stdout.strip()

virt-libguestfs: Fix bugs in virt_sysprep test

virt: propcan clean some old style construction V2

workaround for ovs

Signed-off-by: NLi Yang <liyang.fnst@cn.fujitsu.com>

It is possible to call 'virsh setmem --domain test 524288' but the current
code create 'virsh setmem 524288 --domain test' for the right parameters.
Signed-off-by: NPavel Hrdina <phrdina@redhat.com>

The domain name and memory size have to be defined to PASS the test
on newer libvirt.

There are some tests that have different result on older libvirt and
newer libvirt. The right condition is that the new libvirt should not
fail, but the older one should fail. By this patch we follow this
behaviour.
Signed-off-by: NPavel Hrdina <phrdina@redhat.com>

We should restore memory into original size by using the original_outside_mem
instead of original_inside_mem because the inside mem is lower than the outside
mem and because of the virsh setmem command is setting the outside mem.
Signed-off-by: NPavel Hrdina <phrdina@redhat.com>

Signed-off-by: NDongsheng Yang <yangds.fnst@cn.fujitsu.com>

Nettype user default fixes

Signed-off-by: NChris Evich <cevich@redhat.com>

This change mirrors the default as used by the test runner.  I
added/rearranged the comments to make it clear it will be overridden.
Signed-off-by: NChris Evich <cevich@redhat.com>

Sample output:

15:37:08 INFO | Starting test ...
15:37:08 DEBUG| Test parameters:
                     ...
15:39:52 WARNI|
15:39:52 WARNI| Testing with nettype='user' while running as root may
                produce unexpected results!!!
15:39:52 WARNI|
Signed-off-by: NChris Evich <cevich@redhat.com>

When test runner is executed as root, default to using bridge-mode
networking.  Otherwise, force it to user-mode, unless different
on the command line.
Signed-off-by: NChris Evich <cevich@redhat.com>

Include more details about the NIC and Network in question when
reporting possible port-redirection problems.  Sample:

14:57:16 DEBUG| Destroying VM
14:57:16 DEBUG| Trying to shutdown VM with shell command
14:57:16 DEBUG| Don't know how to connect to guest port 22 with
                networking type 'user', to destination 'virbr0',
                for nic 'nic1' with mac '52:54:00:5d:5e:5f' and
                ip 'None'.
Signed-off-by: NChris Evich <cevich@redhat.com>

virttest.qemu_devices: Workaround unsupported -drive addr=...

Antient versions of qemu don't support -drive addr=... option.
This workaround logs and ignores setting of addr.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Turns out staging also needed __init__.py fixing. Sorry
about that once again.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Due to a very bad mistake of mine, I hose the ability
to access utils_test.qemu and utils_test.libvirt. By
importing libvirt and qemu internally those problems
should be fixed. Sorry guys.
Signed-off-by: NLucas Meneghel Rodrigues <lmr@redhat.com>

Currently, when close tap device, it's can not automatically cleaned
from openvswitch-db, sinece openvswitch have it's own userspace db, and
that ovsdb is considered to be only modified by the user. when
qemu exit by error, the ovs will never delete the port from it's
db, even kernel have destroied the port, as a result, when we run a
loop using ovs, it alway raise error for the port already exist.

This patch just add a workaround, make the test loop using ovs
can run normally.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Now when using find_current_bridge, if the host not install openvswitch
it will raise like:
AttributeError: 'NoneType' object has no attribute 'port_to_br'

This patch add a check whether ovs is exist before using it.
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

Whitespace character("\n") exist in the stdout data, when run
'ovs-vsctl' command in some system. better remove it before return
Signed-off-by: NYunping Zheng <yunzheng@redhat.com>

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

Conflicts:
	qemu/cfg/multi-host-tests.cfg
	virttest/utils_test.py