It was inheriting from Exception, so the rescue clause in
external_feed_aggregator.rb wasn't catching it, causing the entire
ExternalFeedAggregator.process job to fail.
Looking at the other uses of CanvasHttp, I see no reason to believe that
this was intentional.
closes CNVS-21213
Test plan: as mentioned above the problem here is that when an
ExternalFeed throws this error, it was not being caught properly and so
all external feeds afterwards in the queue would not be processed. In
order to recreate these conditions we need a feed that fails and then we
need to make sure there are more feeds afterwards in the queue.
- First add a valid rss external feed. We'll use this just to make sure
the system is working.
- Then add a url (doesn't even have to be to an rss feed) that redirects
more than 3 times before resolving. This will trigger the redirect
error, that was causing the problem. You can verify if a url will break
by running CanvasHttp.get('http://broken.redirects') in a rails console
- Then add another valid rss feed. This will be the one that comes
after the broken one, and it working successfully is what we're
looking for.
- The system loads feeds to check in order of a "refresh_date" which is
initially set to the time they were created, and then reset to the last
time they were updated. This is why the order of creation is important
above, because we'll create a refresh order that has a valid feed after
a broken feed.
- The feed import system is set up to ignore rss entries that were
created before the feed object was created. To get around this, you
can manually update the created_at time of these objects. Find a time
that is before the last entry in the two valid feeds, and run:
`ExternalFeed.update_all(created_at: 6.hours.ago)`
- Finally we're set, just run this in the console, to simulate the
delayed job that runs:
`ExternalFeedAggregator.process`
- And ensure that new entries from both the first and third feeds are
imported.
Change-Id: If60158719d1b618c1e706ea0d056bc6b08f87457
Reviewed-on: https://gerrit.instructure.com/53494
Reviewed-by: Simon Williams <simon@instructure.com>
Tested-by: Jenkins
QA-Review: Jeremy Putnam <jeremyp@instructure.com>
Product-Review: Simon Williams <simon@instructure.com>
554789efbf