[Libguestfs] [PATCH] rhv-upload: Fix waiting for transfer

[Nir Soffer]

On Thu, Nov 28, 2019 at 10:58 PM Richard W.M. Jones <rjones at redhat.com> wrote:
>
> On Thu, Nov 28, 2019 at 09:34:18PM +0200, Nir Soffer wrote:
> > We were not considering failures while initializing the transfer. In
> > this case the transfer phase can change to PAUSED_SYSTEM or
> > FINISHED_FAILURE, and transfer_url will be None, which failed the
> > upload with a misleading error:
> >
> >     RuntimeError: direct upload to host not supported, requires
> >     ovirt-engine >= 4.2 and only works when virt-v2v is run within the
> >     oVirt/RHV environment, eg. on an oVirt node
> >
> > Change the wait loop to consider all cases:
> > - Transfer failed and was removed
> > - Transfer failed and will be removed soon
> > - Transfer paused by the system (cancel required)
> > - Unexpected transfer phase (cancel required)
> > - Timeout waiting for TRANSFERRING state (cancel required)
> >
> > Reported-by: Xiaodai Wang
> > ---
> >
> > I could easy simulate the case when the system paused the transfer by
> > injecting an error in vdsm, failing transfer initialization.
> >
> > The import fail with:
> >
> > nbdkit: python[1]: error: /home/nsoffer/src/virt-v2v/tmp/rhvupload.1DgXyh/rhv-upload-plugin.py: open: error: Traceback (most recent call last):
> >    File "/home/nsoffer/src/virt-v2v/tmp/rhvupload.1DgXyh/rhv-upload-plugin.py", line 109, in open
> >     transfer = create_transfer(connection, disk, host)
> >    File "/home/nsoffer/src/virt-v2v/tmp/rhvupload.1DgXyh/rhv-upload-plugin.py", line 539, in create_transfer
> >     "transfer %s was paused by system" % transfer.id)
> >  RuntimeError: transfer 32b97384-ac8b-40d5-b423-26d31faabe32 was paused by system
> >
> > I could not simulate the other cases. This probaly requires injecting
> > errors in engine.
>
> You might be able to inject errors more easily than that by modifying
> the test harness (tests/test-v2v-o-rhv-upload-module/ovirtsdk4/).
>
> Anyway patch looks reasonable, although I didn't test it, so:
>
> ACK

Seems that Daniel is too busy now to review this, so I think we should
push this.