I’m trying to set up the tables for a new django project (that is, the tables do NOT already exist in the database); the django version is 1.7 and the db back end is PostgreSQL. The name of the project is crud. Results of migration attempt follow:

python manage.py makemigrations crud

Migrations for 'crud':
    - Create model AddressPoint
    - Create model CrudPermission
    - Create model CrudUser
    - Create model LDAPGroup
    - Create model LogEntry
    - Add field ldap_groups to cruduser
    - Alter unique_together for crudpermission (1 constraint(s))

python manage.py migrate crud

Operations to perform:
  Apply all migrations: crud
Running migrations:
  Applying crud.0001_initial...Traceback (most recent call last):
  File "manage.py", line 18, in <module>
  File "/usr/local/lib/python2.7/dist-packages/django/core/management/__init__.py", line 385, in execute_from_command_line
  File "/usr/local/lib/python2.7/dist-packages/django/core/management/__init__.py", line 377, in execute
  File "/usr/local/lib/python2.7/dist-packages/django/core/management/base.py", line 288, in run_from_argv
    self.execute(*args, **options.__dict__)
  File "/usr/local/lib/python2.7/dist-packages/django/core/management/base.py", line 338, in execute
    output = self.handle(*args, **options)
  File "/usr/local/lib/python2.7/dist-packages/django/core/management/commands/migrate.py", line 161, in handle
    executor.migrate(targets, plan, fake=options.get("fake", False))
  File "/usr/local/lib/python2.7/dist-packages/django/db/migrations/executor.py", line 68, in migrate
    self.apply_migration(migration, fake=fake)
  File "/usr/local/lib/python2.7/dist-packages/django/db/migrations/executor.py", line 102, in apply_migration
    migration.apply(project_state, schema_editor)
  File "/usr/local/lib/python2.7/dist-packages/django/db/migrations/migration.py", line 108, in apply
    operation.database_forwards(self.app_label, schema_editor, project_state, new_state)
  File "/usr/local/lib/python2.7/dist-packages/django/db/migrations/operations/models.py", line 36, in database_forwards
  File "/usr/local/lib/python2.7/dist-packages/django/db/backends/schema.py", line 262, in create_model
    self.execute(sql, params)
  File "/usr/local/lib/python2.7/dist-packages/django/db/backends/schema.py", line 103, in execute
    cursor.execute(sql, params)
  File "/usr/local/lib/python2.7/dist-packages/django/db/backends/utils.py", line 82, in execute
    return super(CursorDebugWrapper, self).execute(sql, params)
  File "/usr/local/lib/python2.7/dist-packages/django/db/backends/utils.py", line 66, in execute
    return self.cursor.execute(sql, params)
  File "/usr/local/lib/python2.7/dist-packages/django/db/utils.py", line 94, in __exit__
    six.reraise(dj_exc_type, dj_exc_value, traceback)
  File "/usr/local/lib/python2.7/dist-packages/django/db/backends/utils.py", line 66, in execute
    return self.cursor.execute(sql, params)
django.db.utils.ProgrammingError: relation "crud_crudpermission" already exists

Some highlights from the migration file:

dependencies = [
    ('auth', '0001_initial'),
    ('contenttypes', '0001_initial'),
            ('id', models.AutoField(verbose_name="ID", serialize=False, auto_created=True, primary_key=True)),
            ('_created_by', models.CharField(default=b'', max_length=64, null=True, editable=False, blank=True)),
            ('_last_updated_by', models.CharField(default=b'', max_length=64, null=True, editable=False, blank=True)),
            ('_created', models.DateTimeField(null=True, editable=False, blank=True)),
            ('_last_updated', models.DateTimeField(null=True, editable=False, blank=True)),
            ('domain', models.CharField(max_length=32, choices=[(b'town', b'Town'), (b'boe', b'BOE'), (b'police', b'Police')])),
            ('ldap_group', models.CharField(max_length=128, verbose_name=b'LDAP group')),
            ('can_add', models.BooleanField(default=False, verbose_name=b'add')),
            ('can_change', models.BooleanField(default=False, verbose_name=b'change')),
            ('restrict_change_to_own', models.BooleanField(default=False)),
            ('can_delete', models.BooleanField(default=False, verbose_name=b'delete')),
            ('restrict_delete_to_own', models.BooleanField(default=False)),
            ('models', models.ManyToManyField(to='contenttypes.ContentType', null=True, blank=True)),
            'verbose_name': 'CRUD permission',
        unique_together=set([('ldap_group', 'can_add', 'can_change', 'can_delete', 'domain')]),


The crud app is not meant to actually do anything, but I use it another app, so when I try migrate from that app, I trigger the above problem.

I’ve found other examples on the web of people with similar issues, but none of their cases seem to apply because

  1. The problem affects an entire relation, not just one column
  2. I am not using multiple inheritance.

Where should I look next to find the underlying problem?

This works pretty fine

./manage.py migrate --fake default


I’ve faced similar issue when added couple new fields to existing model. I’m using Django 1.9, which introduced --run-syncdb option. Running manage.py migrate --run-syncdb fixed my tables.

Now (I’m using Django 1.9) you can make:

./manage.py migrate [--database DATABASE] --fake [app_label] [migration_name]

This way you’re targeting the problem with more accuracy, and you can fake only the problematic migration on the specific database.

So, looking at the question, you could:

./manage.py migrate --database default --fake crud crud.0001_initial

Been facing a similar issue, eventually deleted all .py files in migration folder (django 1.7 creates one automatically), worked perfectly after that.

I was facing the similar issues, where i had changed column name. I was getting same error as mentioned in the stack-trace provided with he question.

Here’s what I did.

I ran fake migrations first. Then i removed it’s(migrations i wanted to run) entry from django_migrations table and ran migrations again(no fake this time).

Changes appeared as expected for me.

hope this is helpful.

Django provides a --fake-initial option which I found effective for my use. From the Django Migration Documentation:


Allows Django to skip an app’s initial migration if all database
tables with the names of all models created by all CreateModel
operations in that migration already exist. This option is intended
for use when first running migrations against a database that
preexisted the use of migrations. This option does not, however, check
for matching database schema beyond matching table names and so is
only safe to use if you are confident that your existing schema
matches what is recorded in your initial migration.

For my use, I had just pulled a project from version control and was preparing to add some new model fields. I added the fields, ran ./manage.py makemigrations and then attempted to run ./manage.py migrate which threw the error since, as one would expect, many of the fields already existed in the existing database.

What I should have done was to run makemigrations immediately after pulling the project from versions control to create a snapshot of existing models’ state. Then, running the ./manage.py migrate --fake-initial would be the next step.

After that you can add away and makemigrations > migrate as normal.

NOTE: I do not know if a --fake-initial would skip existing fields and add new ones. I opted to comment out the new fields I’d created up to that point, run the --fake-initial as if it were the first thing I did after pulling from version control, then added in updated fields in the next migration.

Other related documentation: https://docs.djangoproject.com/en/dev/topics/migrations/#initial-migrations

I’ve been dealing with this for several years.
There could be different scenarios:

Scenario 1: as in the original post, you had no tables to start with.
In this case, I’d

  1. comment out the relationship in models.py
  2. run python manage.py
  3. migrate assuming that it now succeeds
  4. uncomment what you
  5. commented out in step 1 run python manage.py migrate –fake

Scenario 2: multiple apps:
One possibility is that you might have different apps and the data model of one app is using some tables from the other app. In this case, if the data model is designed properly, you should be able to create the tables for one app only (by specifying only that one in setting.py), then add the other app and migrate. If it is not design with care, and there are recursive dependencies, I suggest changing the design rather than making a temporary fix.

Scenario 3: you had some tables and something went wrong with your migration, then I’d

  1. revert models.py to what it was and only introduce the new
    relationship that appears to already exist in models.py.
  2. delete migration folder
  3. run python manage.py makemigrations
  4. introduce new changes to models.py if any and continue with makemigrations and
    migrate commands as usual.

I found and solved a particular example of this error in a Django 1.10 project while I was changing a foreign key field named member to point to a different table. I was changing this in three different models and I hit this error on all of them. In my first attempt I renamed member to member_user and tried to create a new field member as a foreign key pointing at the new table, but this didn’t work.

What I found is that when I renamed the member column it did not modify the index name in the form <app>_<model>_<hash> and when I tried to create a new member column it tried to create the same index name because the hash portion of the name was the same.

I resolved the problem by creating a new member_user relation temporarily and copying the data. This created a new index with a different hash. I then deleted member and recreated it pointing at the new table and with it the would-be conflicting index name. Once that was done I ran the RunPython step to populate the new member column with references to the applicable table. I finished by adding RemoveField migrations to clean up the temporary member_user columns.

I did have to split my migrations into two files because I received this error:

psycopg2.OperationalError: cannot ALTER TABLE “<table_name>” because it has pending trigger events

After the creation and copy of data into member_user I was not able to remove member in the same migration transaction. This may be a postgres specific limitation, but it was easily resolved by creating another transaction and moving everything after creating and copying member_user into the second migration.

In my case, a migration file was deleted and a new one was auto-generated, which had a different name. Because of the name difference, Django tried to apply the new migration file, which was exactly same as the previously applied one, which was now removed. In both of them, a new model had to be created which resulted in django.db.utils.ProgrammingError: relation "app_space" already exists. I tried to reverse the migration, but the missing migration file prevented django from actually reversing it. Lesson learnt, migration files should be checked into git.

Here are some steps that helped me to get to the bottom of this. --fake solved it temporarily but the same issue happened in the next migration. I wouldn’t recommend --fake unless you know for sure it is the right use case for it.

This answer was really key for me.

  1. Check previous migrations python manage.py showmigrations

  2. Check applied migrations in Django DB select * from django_migrations; (use psql to access postgres db console: psql -h -U <your-db-user> then use target db \c <your-db-name>).

  3. I saw an applied migration that was no longer in my migrations folder.

 20 | app             | 0001_initial                             | 2021-03-05 07:40:20.14602+00
 21 | app             | 0002_some_auto_name                      | 07:40:20.269775+00
 22 | app             | 0003_auto_20210318_2350 <---here         | 2021-03-18 23:50:09.064971+00

But in migrations folder I had 0003_auto_20210318_2355, the same file generated 5 minutes later. I renamed the migration file to the name above so that I could reverse it.

  1. Reverse the migration by passing the migration to which you want to return.

python manage.py migrate <app-name> <latest-migration-to-which-to-return>

python manage.py migrate app 0002_some_auto_name

  1. Do the right thing from here and check in the migrations to git. Then you can do makemigrations and migrate and have a more peaceful life.

For me, When I faced this exception, I solve it using the Django dbshell utility or any kind of MY_DATABASE Viewer / interactive command line.


  1. python manage.py dbshell
  2. ALTER TABLE [name_of_field_that_already_exists] DROP column [field_table];

I found this problem in web2pyframework in models/config.py .


settings.base.migrate = True

on config file to

settings.base.migrate = False

Problem solved.

I recently had the same issue and tried some of the solutions here. manage.py migrate --fake led to a "django_content_type" does not exist error. Equally deleting old migrations might cause problems for other users if the migrations are shared.

The manage.py squashmigrations command (docs) seems to be the ideal way to deal with this. Condenses old migrations into a single migration (which prevents applying them out of sequence etc), and preserves the old migrations for any other users. It worked in my case at least.

I am not sure about using the solution with fake. Most likely that error will occur again at the next migration.

Find out which columns are creating that problem

  1. python manage.py dbshell
  2. select * from <tablename> where false;
    (Now you see which columns are saved by postgresql and can delete them in the database)
  3. alter table <tablename> drop column <columnname>;
    (Start the migrations process)
  4. python manage.py makemigrations
  5. python manage.py migrate

If you’re getting this error when you run python manage.py test --k, you can fix it by deleting the test database: python manage.py test

Do not try to use --fake, because with this you risk corrupting your database.

Instead, you can backup your current database, recreate it, and then apply backup.

  1. Create backup: pg_dump -F c -v -h localhost <database_name> -f tmp/<pick_a_file_name>.psql

  2. Recreate it: rails db:drop db:create

  3. Restore backup: pg_restore --exit-on-error --verbose --dbname=<database_name> tmp/<pick_a_file_name>.psql