Proposal: Create a staging server that uses production database config that is behind a non-LDAP username/password. Why: 1) To ensure we have run all the necessary db updates in production for collections to launch 2) So that select parties can be given the URL and auth info to create collections that will already exist when we launch Tuesday. Otherwise, we will launch with no collections on the homepage and have to scramble to create collections, favorite them so they surface to the top, and flush the cache. How: 1) We need to ensure running the SQL in production won't break the existing site. So, we will need to check out the current revision in production and a dump of production, and apply the SQL and make sure nothing breaks. 2) File the IT bug for box creation 3) File the IT bug for SQL to be run in production
Would like to have this setup for Monday, but Friday would be even better so I can do screencasts over the weekend. Wil, do you have someone in mind to test the SQL?
(In reply to comment #1) > Would like to have this setup for Monday, but Friday would be even better so I > can do screencasts over the weekend. > > Wil, do you have someone in mind to test the SQL? Not yet. Assuming we found time and disk space (heh) to load up a new db and install another version of AMO, what is your testing strategy? I don't think you can hit all areas of AMO to make sure it still works while we're in transition.
All of the SQL should be collection-related. So we should know where to look based on the SQL that is run. I think any problems would be obvious with the Cake "missing table" or non-existent column errors.
Nah, there is more than just collections: https://wiki.mozilla.org/Update:Developers/Database_Changes#5.0.6 I need to sort through it again since some of it already got run in production but the majority of it hasn't been.
Looking through the 5.0.6 SQL, I see only table and column creation and a couple key additions, with the exception of dropping the unused vanilla columns. So I think we'll be pretty safe. The only thing we might want to play with a little is Fashion Your Firefox.
In theory, sure, but betting the stability of the site on glancing through the queries doesn't seem like a good idea. I'd be more comfortable just pushing like normal and creating the collections afterwards.
wil- we need to figure out a way to get these collections created in production somehow. Otherwise the home page will look terrible as we rush to create them. worst case is we push in the middle of the night, hope nothing goes wrong, and create collections during low traffic- but even then europe will be on AMO. any suggestions on how to get this tested satisfactorily would be a huge help. thanks!
(In reply to comment #7) > wil- we need to figure out a way to get these collections created in production > somehow. Otherwise the home page will look terrible as we rush to create them. > > worst case is we push in the middle of the night, hope nothing goes wrong, and > create collections during low traffic- but even then europe will be on AMO. This is what I'm suggesting. I don't think the front page will look "terrible." Instead of scrolling through collections it'll just show the introduction for a few minutes until they are created.
(In reply to comment #8) > This is what I'm suggesting. I don't think the front page will look > "terrible." Instead of scrolling through collections it'll just show the > introduction for a few minutes until they are created. Agreed: I don't have these collections locally, and it looks like this: http://skitch.com/fredw/b19i3/collections-no-promos
Ok, we can create collections after launch. It will probably take me about 30 minutes to set everything up, after which we'll need to have IT flush the caches and we'll need to get everyone to favorite the collections so they are the most popular for the start.
Status: NEW → RESOLVED
Last Resolved: 10 years ago
Resolution: --- → WONTFIX
Product: addons.mozilla.org → addons.mozilla.org Graveyard
You need to log in before you can comment on or make changes to this bug.