Add 'lago ovirt stop/start'

[nvgoldin]

Fixes: #463

Per what it is supposed to do - it seems to be working.
I tested it against OST 'basic_suite_master', after triggering lago ovirt start the iSCSI storage domain failed to start. On a second round(lago ovirt stop && lago ovirt start), one of the hosts became none-operational and was unable to connect to the storage domain(other host was fine, and the iSCSI domain came back to life too).

[gbenhaim]

Why did you remove this check ?

[nvgoldin]

Because it failed the run, but after a few seconds the host reported 'UP' in the engine.

[gbenhaim]

So maybe add "allowed_exception" in "testlib.assert_true_within" ?

[nvgoldin]

Its not an exception :/ For the good and bad:

In [3]: from ovirtsdk4 import types

In [4]: type(types.HostStatus)
Out[4]: enum.EnumMeta

In [5]: type(types.HostStatus.NON_RESPONSIVE)
Out[5]: <enum 'HostStatus'>

There is some refactoring we can do here, with #483 or without, but I think we can do it later.

[nvgoldin]

Erhm, while testing this I stumbled with this exception from the SDK:

> lago ovirt stop-vms
Error occured, aborting
Traceback (most recent call last):
  File "/home/ngoldin/src/nvgoldin.github.com/lago/ovirtlago/cmd.py", line 310, in do_run
    self.cli_plugins[args.ovirtverb].do_run(args)
  File "/home/ngoldin/src/nvgoldin.github.com/lago/lago/plugins/cli.py", line 184, in do_run
    self._do_run(**vars(args))
  File "/home/ngoldin/src/nvgoldin.github.com/lago/lago/utils.py", line 495, in wrapper
    return func(*args, **kwargs)
  File "/home/ngoldin/src/nvgoldin.github.com/lago/lago/utils.py", line 506, in wrapper
    return func(*args, prefix=prefix, **kwargs)
  File "/home/ngoldin/src/nvgoldin.github.com/lago/ovirtlago/cmd.py", line 197, in do_ovirt_stop_vms
    prefix.virt_env.engine_vm().stop_all_vms()
  File "/home/ngoldin/src/nvgoldin.github.com/lago/ovirtlago/virt.py", line 277, in stop_all_vms
    [vms_service.vm_service(id).stop() for id in ids]
  File "/home/ngoldin/virtualenv/lago-venv/lib/python2.7/site-packages/ovirtsdk4/services.py", line 30209, in stop
    self._check_action(response)
  File "/home/ngoldin/virtualenv/lago-venv/lib/python2.7/site-packages/ovirtsdk4/service.py", line 129, in _check_action
    Service._raise_error(response, result.fault)
  File "/home/ngoldin/virtualenv/lago-venv/lib/python2.7/site-packages/ovirtsdk4/service.py", line 71, in _raise_error
    raise Error(msg)
Error: Fault reason is "Operation Failed". Fault detail is "[Virtual machine destroy error]". HTTP response code is 400.

But the VM was actually stopped successfully on the engine side.

While looking at the Ansible module earlier, I saw:
https://github.com/ansible/ansible/blob/devel/lib/ansible/module_utils/ovirt.py#L271

def get_entity(service):
    """
    Ignore SDK Error in case of getting an entity from service.
    """
    entity = None
    try:
        entity = service.get()
    except sdk.Error:
        # We can get here 404, we should ignore it, in case
        # of removing entity for example.
        pass
    return entity

@machacekondra - Any idea if this is an expected behaviour?

Maybe the nested call is the issue? (vms_service.vm_service(id).stop() )

[machacekondra]

This is some backend error, not Python SDK. Feel free to open an bug on it.

[nvgoldin]

@machacekondra - I see, thanks.

I removed the start L2 VMs from lago ovirt start, as this is not fully baked yet: I suspect we need to first assert all storage domains are in active mode, and only then start the VMs.

@gbenhaim - this is ready on my side.

[gbenhaim]

ci merge please