Closed giorgiosironi closed 6 years ago
Seem __pycache__
directory is being commited
I'll be going on implementing feedback on Wednesday, assuming the general framework of #28 is confirmed.
Integrated with article store from master
(only issue the missing runId
), will also check the dashboard after it has changed to the new format.
Merged master
, but the article store doesn't start (no logs are written) as requests to the container's 8000
port are refused despite the python manage.py runserver
process being running.
It starts when running with docker-compose up article-store
, but docker-compose up -V
reproduces the problem for me.
Non-deterministic problem.
Some manual tinkering lead to:
Run 4df2280c-67f8-11e8-9208-0242ac1a0010
event.article.version.post.started
Jun 4, 2018, 1:07:59 PM
event.downstream.crossref.started
We are happy to receive this paper and have put it into a queue
Jun 4, 2018, 1:07:59 PM
event.article.version.post.completed
Jun 4, 2018, 1:07:59 PM
event.downstream.crossref.started
We are happy to receive this paper and have put it into a queue
Jun 4, 2018, 1:07:59 PM
Had to manually bind the queue to the second exchange however: https://github.com/php-amqplib/RabbitMqBundle/issues/367
Thought about adding depends_on: - downstream-sample
for the dashboard-events
, but it doesn't have that for the articles
exchange and it's just a startup order (or build time) dependency, so wouldn't add any value.
Can see it working, but seems to respond to any article event? I see two downstream.crossref.started
events: one after article.version.post.started
and one after article.version.post.completed
. Would only expect to see the latter.
Successfully receiving the downstream events locally.
Are we going to have some sort of filtering on the received events here? e.g. only act on type article.version.*.completed
or something along those lines? I know its not strictly important for demo purposes but with its current implementation, once you add in the work flow steps it will be firing off downstream.crossref.started
about 10+ times when realistically you would only want to trigger downstream once the initial article 'processing' has completed.
I'll put a filter on article.version.post.completed
that seems the correct integration point.
Took a bit to understand the flow of routing keys but :+1: feel free to merge when conveniente.
For #29
See README for how much is implemented in this PR and what will follow.
Sample log: