While developing a Django project tracking it with git and GitHub, how should I manage migrations?
Sometimes when I deploy a release to production some migrations crash due to files that I delete after this migration.
How can I avoid this?
Thanks.
There is other threads on this but basically this is the rules I use:
You should definately remote migrations files using Git.
Never run makemigrations on production environment always in developpment.
Now, let's say you made a change on one of your models (in developpment I hope), you will run a normal makemigrations. Then, run migrate (still in dev) in order to test everything. When you're ready, you will commit and push the created files and pull in prod to then run migrate to update database schema.
This will assure good versionning of your migrations files. Also, it will greatly help you in the long run, because running makemigrations in produciton and in dev simultaneously will just cause more conflicts on migrations files which can be a pain.
I've been trying to wrap my head around how to avoid creating a mess with Django migrations. I tried "don't delete migrations" and for the most part that worked, but when I wanted to delete an app off my codebase, I struggled and fell back into a mess.
I was thinking, it would be far easier to untangle these issue if one could:
1- delete all current migrations
2- create a migration from an existing database schema (rather than from Django models) <- this is the missing step
3- run migrate --fake to synch DB with migrations
4- run makemigrations to add any non-applied changes to the migration chain (this time from Django models)
5- run migrate to synch the DB with the new changes
The only step that I don't know how to do is step 2. Is this doable? I don't know of any built in module or tool that does this so I'm wondering why not.
Some research showed python manage.py inspectDB gets me partially through step 2. But is there an automated way to infer migrations from an existing database?
I've been working on a contract project in Django, and I performed some updates for the live server, including updates that went from South to an up to date Django version.
I was attempting to follow the instructions in the documentation, which states to delete all the old migration files except init, and then use makemigrations and "python manage.py migrate --fake-initial".
But it turns out database permissions were screwy. I was not able to apply the update.
The server owner (who was the past contractor) did not yet realize that there was a permission issue, and tried the update on his own.
The server is now working, but he stated to me he "had to do many changes to the migrations working because there was no old migrations".
I am assuming he did not know the proper way to upgrade from South to integration migrations with Django, and did it all manually rather than letting Django migrations do the work.
Well now I am ready for future updates, but I look in the migrations folder, and it looks like both the old migrations files as well as the new migrations files are in there? An example: http://screencast.com/t/gYQdc1tl4 (notice the dates).
Everything seems to be working, but I am wondering... how should I go about future migrations from here? Will a regular makemigrations/migrate work? Or should I do the --fake-initial again?
Any suggestions please? Since we strayed from documentation I'm unsure of how to go about this next.
Thanks
If everything works now and you don't need the migrations for your development server, then I would just remove all old migrations (do a backup!) and start with "new migrations" using --fake-initial
Should I be adding the Django migration files in the .gitignore file?
I've recently been getting a lot of git issues due to migration conflicts and was wondering if I should be marking migration files as ignore.
If so, how would I go about adding all of the migrations that I have in my apps, and adding them to the .gitignore file?
Quoting from the Django migrations documentation:
The migration files for each app live in a “migrations” directory inside of that app, and are designed to be committed to, and distributed as part of, its codebase. You should be making them once on your development machine and then running the same migrations on your colleagues’ machines, your staging machines, and eventually your production machines.
If you follow this process, you shouldn't be getting any merge conflicts in the migration files.
When merging version control branches, you still may encounter a situation where you have multiple migrations based on the same parent migration, e.g. if to different developers introduced a migration concurrently. One way of resolving this situation is to introduce a merge_migration. Often this can be done automatically with the command
./manage.py makemigrations --merge
which will introduce a new migration that depends on all current head migrations. Of course this only works when there is no conflict between the head migrations, in which case you will have to resolve the problem manually.
Given that some people here suggested that you shouldn't commit your migrations to version control, I'd like to expand on the reasons why you actually should do so.
First, you need a record of the migrations applied to your production systems. If you deploy changes to production and want to migrate the database, you need a description of the current state. You can create a separate backup of the migrations applied to each production database, but this seems unnecessarily cumbersome.
Second, migrations often contain custom, handwritten code. It's not always possible to automatically generate them with ./manage.py makemigrations.
Third, migrations should be included in code review. They are significant changes to your production system, and there are lots of things that can go wrong with them.
So in short, if you care about your production data, please check your migrations into version control.
You can follow the below process.
You can run makemigrations locally and this creates the migration file. Commit this new migration file to repo.
In my opinion you should not run makemigrations in production at all. You can run migrate in production and you will see the migrations are applied from the migration file that you committed from local. This way you can avoid all conflicts.
IN LOCAL ENV, to create the migration files,
python manage.py makemigrations
python manage.py migrate
Now commit these newly created files, something like below.
git add app/migrations/...
git commit -m 'add migration files' app/migrations/...
IN PRODUCTION ENV, run only the below command.
python manage.py migrate
Quote from the 2022 docs, Django 4.0. (two separate commands = makemigrations and migrate)
The reason that there are separate commands to make and apply
migrations is because you’ll commit migrations to your version control
system and ship them with your app; they not only make your
development easier, they’re also useable by other developers and in
production.
https://docs.djangoproject.com/en/4.0/intro/tutorial02/
TL;DR: commit migrations, resolve migration conflicts, adjust your git workflow.
Feels like you'd need to adjust your git workflow, instead of ignoring conflicts.
Ideally, every new feature is developed in a different branch, and merged back with a pull request.
PRs cannot be merged if there's a conflict, therefore who needs to merge his feature needs to resolve the conflict, migrations included. This might need coordination between different teams.
It is important though to commit migration files! If a conflict arises, Django might even help you solve those conflicts ;)
I can't imagine why you would be getting conflicts, unless you're editing the migrations somehow? That usually ends badly - if someone misses some intermediate commits then they won't be upgrading from the correct version, and their copy of the database will be corrupted.
The process that I follow is pretty simple - whenever you change the models for an app, you also commit a migration, and then that migration doesn't change - if you need something different in the model, then you change the model and commit a new migration alongside your changes.
In greenfield projects, you can often delete the migrations and start over from scratch with a 0001_ migration when you release, but if you have production code, then you can't (though you can squash migrations down into one).
The solution usually used, is that, before anything is merged into master, the developer must pull any remote changes. If there's a conflict in migration versions, he should rename his local migration (the remote one has been run by other devs, and, potentially, in production), to N+1.
During development it might be okay to just not-commit migrations (don't add an ignore though, just don't add them). But once you've gone into production, you'll need them in order to keep the schema in sync with model changes.
You then need to edit the file, and change the dependencies to the latest remote version.
This works for Django migrations, as well as other similar apps (sqlalchemy+alembic, RoR, etc).
Gitignore the migrations, if You have separate DBs for Development, Staging and Production environment. For dev. purposes You can use local sqlite DB and play with migrations locally.
I would recommend You to create four additional branches:
Master - Clean fresh code without migrations. Nobody is connected to this branch. Used for code reviews only
Development - daily development. Push/pull accepted. Each developer is working on sqlite DB
Cloud_DEV_env - remote cloud/server DEV environment. Pull only. Keep migrations locally on machine, which is used for the code deployment and remote migrations of Dev database
Cloud_STAG_env - remote cloud/server STAG environment. Pull only. Keep migrations locally on machine, which is used for the code deployment and remote migrations of Stag database
Cloud_PROD_env - remote cloud/server DEV environment. Pull only. Keep migrations locally on machine, which is used for the code deployment and remote migrations of Prod database
Notes:
2, 3, 4 - migrations can be kept in repos but there should be strict rules of pull requests merging, so we decided to find a person, responsible for deployments, so the only guy who has all the migration files - our deploy-er. He keeps the remote DB migrations each time we have any changes in Models.
You should think of migrations as a version control system for your database schema. makemigrations is responsible for packaging up your model changes into individual migration files - analogous to commits - and migrate is responsible for applying those to your database.
The migration files for each app live in a “migrations” directory inside of that app, and are designed to be committed to, and distributed as part of, its codebase. You should be making them once on your development machine and then running the same migrations on your colleagues’ machines, your staging machines, and eventually your production machines.
golden rule : Make once on dev and migrate on all
Having a bunch of migration files in git is messy. There is only one file in migration folder that you should not ignore. That file is init.py file, If you ignore it, python will no longer look for submodules inside the directory, so any attempts to import the modules will fail. So the question should be how to ignore all migration files but init.py?
The solution is:
Add '0*.py' to .gitignore files and it does the job perfectly.
Hope this helps someone.
Committing your migrations is just a recipe for disaster. Because the migrations are somewhat or a chain that can be traced back, if you have dependences from a former migration e.g a pip module which you used at some point in your project lifecycle and then stopped using. You might find bread crumbs of such dependences in your migrations thread and you have to manually remove these imports from the migrations file.
Verdict, except you are a god tier Django dev, probably avoid adding migrations to your commits.
Short answer
I propose excluding migrations in the repo. After code merge, just run ./manage.py makemigrations and you are all set.
Long answer
I don't think you should put migrations files into repo. It will spoil the migration states in other person's dev environment and other prod and stage environment. (refer to Sugar Tang's comment for examples).
In my point of view, the purpose of Django migrations is to find gaps between previous model states and new model states, and then serialise the gap. If your model changes after code merge, you can simple do makemigrations to find out the gap. Why do you want to manually and carefully merge other migrations when you can achieve the same automatically and bug free? Django documentation says,
They*(migrations)*’re designed to be mostly automatic
; please keep it that way. To merge migrations manually, you have to fully understand what others have changed and any dependence of the changes. That's a lot of overhead and error prone. So tracking models file is sufficient.
It is a good topic on the workflow. I am open to other options.
I want to revert my last migration (0157) by running its Migration.backwards() method. Since I am reverting the migration in production server I want to run it automatically during code deployment. Deployment script executes these steps:
Pull code changes
Run migrations: manage.py migrate <app>
Refresh Apache to use newest code: touch django.wsgi
If I could, I would create new migration file which would tell South to backward migrate to 0156:
migrations/0158_backward__migrate_to_0156.py
This commited migration would be deployed to production and executed during manage.py migrate <app> command. In this case I wouldn't have to execute backward migration by hand, like suggested in these answers.
Lets say, I have created two data migrations, first for user's Payment, second for User model. I have implemented backwards() methods for both migrations in case I'd have to revert these data migrations. I've deployed these two migrations to production. And suddenly find out that Payment migration contains an error. I want to revert my two last data migrations as fast as possible. What is the fastest safe way to do it?
Since I am reverting the migration in production server I want to run
it automatically during code deployment.
IMHO the safest path is
run manage.py migrate <app> (i.e. apply all existing migrations, i.e. up to 0156)
undo the changes in your model
run manage.py schemamigration <app> --auto
This will create a new migration 0157 that effectively reverts the previous migration 0156. Then simply apply the new migration by running manage.py migrate <app> again. As I understand, your code deployment will just do that.
Apparently the codeline has migrations up to #157 and now the developer decided that the last one was not a good idea after all. So the plan is to go back to #156.
Two scenarios:
(a) migration #157 was not released or deployed anywhere yet.
Simply revert the last change from models.py and delete migration #157.py from the source archive. Any deployment will take the system to level 156; "157 was never there".
(b) there have been deployments of the latest software with migration #157.
In this case the previous strategy will obviously not work. So you need to create a migration #158 to undo #157. Revert the change in models.py and run
django manage.py migrate <app> 0157
django manage.py schemamigration <app> --auto
This will auto-generate a new migration #158, which will contain the inverse schema migration compared to #157.
If schemamigration is giving trouble because of django Model validation (something that can happen if you have custom validators which check stuff outside the ORM box), I suggest the following workaround:
<django project>/<app>/management/commands/checkmigrations.py
from south.management.commands import schemamigration
class Command(schemamigration.Command):
requires_model_validation = False
help = "schemamigration without model validation"
This command becomes available in manage.py:
django manage.py checkmigrations <app> --auto
There's no silver bullet here. The simplest solution I can think of would be to - in your dev env of course - manually migrate back to 0156, manually update your migration's history table (sorry I can't remember the table's name now) to fool south in thinking you're still #0158, then run schemamigration again. Not garanteed to work but might be worth trying.