Closed saulius closed 11 years ago
@james this quite urgent can we book a deploy slot and reload the db
@matthewford Wrong James I think :smiley:
@matthewford @jabley I'm confused now. Do you need me to do anything here?
Hello!
Sorry about the confusion, I'm blaming the Github mobile app.
I think the tests you've mentioned are valuable. They probably don't belong in smokey - the intention of smokey is a quick 'ping'-like test to confirm that a service is up and serving requests. What we have here is more along the lines of regression / characterisation tests. I think we'll probably create a new repo to put those in. We definitely want to decouple between building / deploying the service versus running regression tests against the service.
We are missing fair amount of national measures on various commodities. This fixes initial CHIEF seed loading: candidate measure's validity end date should have been part of the key. And it used to be, but there was an issue raised about duplicate national measures. Unfortunately, I couldn't find the issue in question, so I can't double check that now. I added a test case so validity end date won't be removed any more. If there will be duplicate measures somewhere we will need to sort that in different way.
And again I'm sorry but the only way to fix this is to reload the database. I uploaded a snapshot snapshot_for-12-12-2012.sql.bz2 do Dropbox. I think we a viable solution based on Event Sourcing to prevent these reloads (at least for the daily updates), that hopefully will be implemented.
I also been digging our previous issues and fixes and come up with a list of things to check for:
Do these qualify as smokey tests? The problem I see is that both Taric and CHIEF may remove any measure at any time, so test will result in error even though nothing bad may have happened.
This list could of course serve as way to check if release was successful.