[Django] #33912: In the admin, saving an object in a non-default database with unique field fails when the default database contains an object with the same unique field value but different PK
Django
fails when the default database contains an object with the same unique
field value but different PK
--------------------------------------------+------------------------
Reporter: François Granade | Owner: nobody
Type: Uncategorized | Status: new
Component: Uncategorized | Version: 4.0
Severity: Normal | Keywords:
Triage Stage: Unreviewed | Has patch: 0
Needs documentation: 0 | Needs tests: 0
Patch needs improvement: 0 | Easy pickings: 0
UI/UX: 0 |
--------------------------------------------+------------------------
Reproducible using the test project in https://github.com/farialima
/django-multidbbug . See README.md on how to run it.
The steps are:
- a model with a unique string field (beside the default int PK)
- two DBs in the settings (`default` and `other`)
- two objects in the default DB with PK 1 and 2, and some unique value for
the unique string field
- a object in the other DB (thus, with PK 1) with a different unique
string field value
- in `admin.py`, point the UI to the other database for the model (using
https://docs.djangoproject.com/en/4.0/topics/db/multi-db/#exposing-
multiple-databases-in-django-s-admin-interface)
- in the admin, try modifying the object in the other DB, giving it the
same unique string value as the object with PK 2 in the default database
=> saving fails with error "Parcel with this Tracking id already exists.",
see screenshot.
The underlying issue is clearly that the admin will try to verify in the
default DB that the objects doesn't exist. I think that the admin should
*not* look in the default DB, as I set it up to only use the 'other' DB
for my model...
```
% poetry run ./manage.py shell
(...)
>>> from multidbbugapp.models import Parcel
>>> print(Parcel.objects.all())
<QuerySet [<Parcel: id=1, tracking_id=one>, <Parcel: id=2,
tracking_id=two>]>
>>> print(Parcel.objects.using('other').all())
<QuerySet [<Parcel: id=1, tracking_id=two-duplicate>]>
>>> o = Parcel.objects.using('other').first()
>>> o
<Parcel: id=1, tracking_id=two-duplicate>
>>> o.tracking_id = 'two'
>>> o.save()
>>>
```
--
Ticket URL: <https://code.djangoproject.com/ticket/33912>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
Django
fails when the default database contains an object with the same unique
field value but different PK
Reporter: François Granade | Owner: nobody
Type: Uncategorized | Status: new
Component: Uncategorized | Version: 4.0
Keywords: | Triage Stage: Unreviewed
Has patch: 0 | Needs documentation: 0
Needs tests: 0 | Patch needs improvement: 0
Easy pickings: 0 | UI/UX: 0
Changes (by François Granade):
* Attachment "Screen Shot 2022-08-10 at 12.07.45 PM.png" added.
screenshot of the error
Django
fails when the default database contains an object with the same unique
field value but different PK
----------------------------------+--------------------------------------
Reporter: François Granade | Owner: nobody
Component: contrib.admin | Version: 4.0
Severity: Normal | Resolution:
Keywords: admin multidb | Triage Stage: Unreviewed
Has patch: 0 | Needs documentation: 0
Needs tests: 0 | Patch needs improvement: 0
Easy pickings: 0 | UI/UX: 0
----------------------------------+--------------------------------------
Changes (by François Granade):
* cc: François Granade (added)
* keywords: => admin multidb
* component: Uncategorized => contrib.admin
* type: Uncategorized => Bug
Old description:
New description:
Reproducible using the test project in https://github.com/farialima
/django-multidbbug . See README.md on how to run it.
The steps are:
- a model with a unique string field (beside the default int PK)
- two DBs in the settings (`default` and `other`)
- two objects in the default DB with PK 1 and 2, and some unique value for
the unique string field
- a object in the other DB (thus, with PK 1) with a different unique
string field value
- in `admin.py`, point the UI to the other database for the model (using
https://docs.djangoproject.com/en/4.0/topics/db/multi-db/#exposing-
multiple-databases-in-django-s-admin-interface)
- in the admin, try modifying the object in the other DB, giving it the
same unique string value as the object with PK 2 in the default database
=> saving fails with error "Parcel with this Tracking id already exists.",
see screenshot.
The underlying issue is clearly that the admin will try to verify in the
default DB that the objects doesn't exist. I think that the admin should
*not* look in the default DB, as I set it up to only use the 'other' DB
for my model...
The issue only occurs in the admin, things are OK when editing the model
objects in code:
{{{
% poetry run ./manage.py shell
(...)
>>> from multidbbugapp.models import Parcel
>>> print(Parcel.objects.all())
<QuerySet [<Parcel: id=1, tracking_id=one>, <Parcel: id=2,
tracking_id=two>]>
>>> print(Parcel.objects.using('other').all())
<QuerySet [<Parcel: id=1, tracking_id=two-duplicate>]>
>>> o = Parcel.objects.using('other').first()
>>> o
<Parcel: id=1, tracking_id=two-duplicate>
>>> o.tracking_id = 'two'
>>> o.save()
>>>
}}}
--
--
Ticket URL: <https://code.djangoproject.com/ticket/33912#comment:1>