Open GoogleCodeExporter opened 9 years ago
Looking at the code, it seems that the real error should have been logged. Did
you see anything in the application logs?
Original comment by riccardo.murri@gmail.com
on 8 Jun 2015 at 3:34
Can you please test with SVN r4283 ?
Original comment by riccardo.murri@gmail.com
on 3 Jul 2015 at 10:20
I'm still seeing this...
File "easybuild/tools/job/gc3pie.py", line 235, in complete
raise EasyBuildError("Failed to create GC3Pie engine: %s", err)
EasyBuildError: 'Failed to create GC3Pie engine: No resources given to
initialize `gc3libs.core.Core` object!'
I broke my GC3Pie setup by removing the paramiko egg:
$ python -c "import gc3libs; import paramiko"
Traceback (most recent call last):
File "<string>", line 1, in <module>
ImportError: No module named paramiko
Original comment by kenneth....@gmail.com
on 3 Jul 2015 at 2:09
I think that the root question is this: what should the Engine do,
when there are multiple resources configured and some of them error
out?
The approach GC3Pie takes ATM is to try to initialize all resources, and
continue if at least one was correctly built. Therefore you only get an
error at the *end* of the process, if no resource could be init'ed
correctly. That error can only be a "generic" error ("No resources"):
what error would you otherwise report if there are two resources and
each of them fails but in a different way?
Of course, one can argue that GC3Pie should instead "fail fast and
hard": i.e., if *any* resource fails initialization, the whole
configuration process should abort immediately. This is however a big
change from the current policy, so it should for now be optional and
*not the default*.
I have implemented this "fail fast" policy: it is selectable by
setting the environment variable
`GC3PIE_RESOURCE_INIT_ERRORS_ARE_FATAL=yes` or via an additional
argument `resource_errors_are_fatal=True` to `create_engine()`, e.g.::
self._engine = create_engine(*self.config_files,
resource_errors_are_fatal=True)
This is available from SVN r4287 onwards.
Original comment by riccardo.murri@gmail.com
on 3 Jul 2015 at 10:07
Original issue reported on code.google.com by
kenneth....@gmail.com
on 8 Jun 2015 at 3:32