log error when at least one list units request fails

[engelmi]

Relates to:
#567
#568

When list units for all nodes is run and the requests are submitted, it might happen that one of those is failing. Previously, this led to a failure when reading the message for the node where the request failed - which got propagated to the caller.
Now, an additional error log in BlueChi is made and the error code is extracted from the sd_bus_message.

One option to reproduce this is by registering a bluechi-agent, but not providing the list units method.
Previous error message with bluechictl:

$ bluechictl list-units
Failed to issue method call: Failed to create a reply message: No such device or address
Call to manager failed: Permission denied

New message:

$ bluechictl list-units
Failed to issue method call: Failed to create a reply message: Invalid request descriptor
Call to manager failed: Permission denied

Not sure this is better for the client, but sd_bus_message_get_errno retrieves error code 53 (Invalid request descriptor). The log in bluechi-controller provides more info:

msg="Failed to list units for node 'node1': Object does not exist at path “/org/eclipse/bluechi/internal/agent”"

[engelmi]

/packit test

[dougsland]

Looks like not related.

 result, output = ctrl.run_python(os.path.join("python", "monitor.py"))
        if result != 0:
>           raise Exception(output)
E           Exception: E
E           ======================================================================
E           ERROR: test_monitor_multiple_nodes_and_units (__main__.TestMonitorMultipleNodesAndUnits)
E           ----------------------------------------------------------------------
E           Traceback (most recent call last):
E             File "/tmp/pljzilvbdp", line 64, in test_monitor_multiple_nodes_and_units
E               assert node_foo.start_unit(service_simple, "replace") != ""
E             File "/usr/lib/python3.9/site-packages/bluechi/api.py", line 574, in start_unit
E               return self.get_proxy().StartUnit(
E             File "/usr/lib/python3.9/site-packages/dasbus/client/handler.py", line 444, in _call_method
E               return self._get_method_reply(
E             File "/usr/lib/python3.9/site-packages/dasbus/client/handler.py", line 477, in _get_method_reply
E               return self._handle_method_error(error)
E             File "/usr/lib/python3.9/site-packages/dasbus/client/handler.py", line 497, in _handle_method_error
E               raise exception from None
E           dasbus.error.DBusError: Method call timed out
E           
E           ----------------------------------------------------------------------
E           Ran 1 test in 30.199s
E           
E           FAILED (errors=1)

[engelmi]

/packit test --label standard

[engelmi]

Looks like not related.
...

Yes, probably the testing farm is under quite a bit of load. Restarted the tests.