I have a deployment pipeline on Heroku which recently started using review apps. This means I have an app - let's call it CI-APP -- which is being created from the master branch.
Every time a pull request is made, a review app is created. We are using Django in our project and so I also added the migrate command to the release phase in the project, so that database migrations can be done automatically.
Today, a coworker submitted a pull request which contained some database changes. The problem is that the migration was ran, and since review apps seem to be using the same database as the app they are suppose to merge to, the migration was applied and now my app CI-APP stopped working...since the code base no longer matches the database structure.
I searched a lot about how to use completely different databases for the review apps compared to the parent app, but to no avail (there are some resources mentioning how you can copy db contents, but that is not what I need).
Any suggestion ?
Update
Ok, so it seems that Heroku does create a new database for the review app, however: the review app copies all of its environment variables from the parent, including the DATABASE_URL (this seems to be the only way to actually create the review app : https://s3.amazonaws.com/heroku-devcenter-files/article-images/1461071037-initial_set_up_review_apps.png)
I think I can do some black magic in the postdeploy script, but since the database generated url can be something such as HEROKU_POSTGRESQL_{color}_URL, I am not sure how to find it ....
To do that, create the app.json file at the root of your project instead of using the heroku dashboard. In this file, you can specify what environment variables to inherit from the parent.
From the heroku docs:
"env": {
"INHERIT_THIS_CONFIG_VAR": {
"required": true
},
"DONT_INHERIT_THIS_CONFIG_VAR": "production"
},
This allows you to specify which database you want to use for the review app. Looking at the documentation of the heroku postgres addon (i assume you're using postgres):
As part of the provisioning process, a DATABASE_URL config var is
added to your app’s configuration. This contains the URL your app uses
to access the database.
So the database_url config variable will be created by the adddon. You simply need to not put it in the app.json file, and it will be created automatically.
Check that you do not have the DATABASE_URL set in the Pipeline Settings CONFIG VARS in Heroku either.. if it is set there, then it seems the Review App will use that as the DB link and not the one created when the Review App is created.
Related
I built a Django project with a few models. Now I created a second server with the same setup. This one is meant to be the deployment server. Databases are separate from the dev-server.
However can you tell me if I can simply copy the databases from the dev server to the deploy or will the Django logic, since I also mean the user models and permissions etc.
The tables which I created myself are no problem to transfer to the new server. However I am wondering If Django gets confused when I also transfer something like the auth_user Model.
Should this work since I also just copied the backend logic as well?
To be more specific, I want to retrieve db setting from a config server when django project starts, and use it to setup django db connection
Someday in the future, the setting in the config server may be changed (for example, change the user password) and pushed to django project then reset the db connection, so I can use new setting without restarting django project or updating project code
Is there a way to do that?
Or what's the right way to hide the db sensitive information (password, etc) from django project code?
Any helps will be grateful, thanks~
So here's my situation...I have two React apps that need to talk to the same Cognito User Pool. I've been able to accomplish this by copying the aws-exports.js file from the first app to the second app I created (not sure if this is something I should be doing or not but it is working). The issue I am having however is when I run an Admin Query on the second app (to say list users in the Cognito User Pool) I get a 403 (Forbidden) error. Has anyone ever run into this before? Googling all day has not helped me so I figured I would ask.
You'll need "multi-frontend" solution:
https://docs.amplify.aws/cli/teams/multi-frontend
I'll give you some useful infos for this:
Open the Amplify Console and there the "first" app (wheres the backend was created).
Go to the first app's "backend" section
Select "Backend environments" tab
Search for "Edit backend" box and this text: "To continue working on the backend, install the Amplify CLI and make updates by running the command below from the root of your project folder"
copy that command, and paste/run in second app's root.
Beware!
do not modify (and push) the backend from the second application.
if you use git branch based environment you must always switch the env AND the branch parallel. Do not pull the "master" backend for your "dev" env.
try to avoid modifing on amplify console if you modify things with amplify cli. Those things cannot be syncronized... :(
If you store multiple apps in a git monorepo:
https://docs.amplify.aws/cli/usage/monorepo
I followed a tutorial from a site and I deployed my django app on pythonanywhere.com.For this purpose I had to use git to upload codes.As a result my settings.py file was accessible by anybody.so any one could see my secret key on settings.py .It is not good right?
I also left DEBUG = True later I found it is not secure.Then I set DEBUG = False on my local machine and it showed it need allowed hosts.What are those things? what should I do? Where can I learn about these?
Given the following architecture of django projects apps and database schemas:
There is a django app which I want to share between several django projects.
Each project has it's own postgresql schema in behind. All schemas live in the same postgresql database.
One project is owner of the apps data, this project is responsible to run migrations and the data should live in its schema.
All other projects may access the data from the other projects schema, because they have a proper postgresql search path set.
We already use this concept with an app that has all models set to unmanaged, which works. But database changes always needs to be done manually. I'd like to benefit from django migrations and therefore I want my models either managed or unmanaged.
Do you think app config is good place to change the models meta? Any other suggestions and approaches on how to solve the requirement are also welcome.
The in the comments mentioned setting MIGRATION_MODULES works very well. In those projects which should only access the data and which should not be responsible for the migrations you can easily disable the migrations like that:
MIGRATION_MODULES = {
'yoursharedapp': None
}
Note that you need to configure the postgresql role used in the those projects with proper grants and search path.