fault source creation failure

[Ben Wyss]

The Fault Source generation function ( https://github.com/gem/oq-ui-api/blob/master/geonode/observations/utils.py#L91 ) needs to include logging and or tests in order to have proper test coverage, and to help identify the source of fault generation failure.

Also as part of this bug, we need to add a more meaning full message to the fault source generation error message that is passed on to the end user in the UI.

Currently the error message reads fault source generated, or fault source generation failed. But it should report why the fault source could not be generated.

[Ben Wyss (bmwyss)]

I have added logging to Django by adding to local_settings.py:

#Logging
ROOT_PATH = os.path.dirname(__file__)
LOG_DIR = "/var/log/geonode/"
LOG_FILE = os.path.join(LOG_DIR, "django.log")

log = logging.getLogger("django")
log.level = logging.DEBUG
formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")

# log to file
fh = logging.FileHandler(LOG_FILE)
fh.setFormatter(formatter)
log.addHandler(fh)

Then in utils.py I added:

import logging
log = logging.getLogger("django.feeds.utils")

and in a function:

...log.debug("d'a fault source name %s" % faultsource.fault)
...

Then, in /var/log/geonode/django.log I can follow the logging for django

[Ben Wyss (bmwyss)]

Postgres is reporting:
could not receive data from client:
Connection reset unexpected EOF on client connection

I will now begin debugging from the client side

[Ben Wyss (bmwyss)]

This means the client is crashing or dropping the connection--one needs
to be looking at the client end for the problem.

[matley (matley)]

The client is crashing as it tries to reuse an existing "jvm" jpype connection. In order to do this you to attach the current process to the current jvm thread. You can find a fix in https://github.com/matley/oq-ui-api/blob/963088/geonode/observations/utils.py

[matley (matley)]

At this moment, the fault creation feature is working on the production server but not on the development server. There is a problem with the environment configuration as the application is not able to start the JVM (it seems that the JAVA_HOME env is not set).

[Matteo Nastasi (nastasi-oq)]

Found where add the the JAVA_HOME set from the working production machine; update the installation script to replicate this modification.