#28676: Using select_for_update with next save() in multiple threads stucks
-------------------------------------+-------------------------------------
Reporter: M1ha Shvn | Owner: nobody
Type: Bug | Status: new
Component: Database layer | Version: 1.10
(models, ORM) |
Severity: Normal | Resolution:
Keywords: django psycopg | Triage Stage:
PostgreSQL select_for_update | Unreviewed
Has patch: 0 | Needs documentation: 0
Needs tests: 0 | Patch needs improvement: 0
Easy pickings: 0 | UI/UX: 0
-------------------------------------+-------------------------------------
Comment (by M1ha Shvn):
Replying to [comment:2 M1ha Shvn]:
> Replying to [comment:1 Tim Graham]:
> > Can you explain why Django is at fault?
>
> 1) According to [https://www.postgresql.org/docs/9.2/static/monitoring-
stats.html PostgreSQL docs] state "idle in transaction" means that query
was executed and control returned to backend code, but it hasn't committed
the transaction yet. But save() is stuck somewhere (it is incorrect). If I
place print(123) after it - it will not print anything. So PostgreSQL
ended its work, but save() method stopped, waiting for something and not
continuing code execution. The expected behavior of django here is to
leave save() method, than leave "with transaction.atomic" context manager
and commit transaction, so other transaction can go in. The proof is
replacing save() method with QuerySet.update(). As you can see on the
screen, it generates perfectly the same SQL, but works fine without
stucking.
> 2) The second strange factor is duplicate select query, generated inside
save() method. As you can see from above and screen, it's the same query
without "FOR UPDATE". But there is no need in this query - conversation
data has been already selected by "SELECT ... FOR UPDATE". According to
[https://docs.djangoproject.com/en/1.10/ref/models/instances/#what-
happens-when-you-save the docs], save() should do only one query - INSERT
or UPDATE (UPDATE in this situation) without any SELECT queries.
> P. s. The only reason for save() of doing second SELECT from the docs is
getting pk value. But it is not INSERT query, pk is already defined.
Moreover force_update=True doesn't change anything.
>
> Replying to [comment:1 Tim Graham]:
> > propose a change to fix it
> I'm not quite good in django inner code, I've tried debugging where,
inside save() method, the problem is, but haven't succeeded
1) Futher investigation shows that [https://github.com/Suor/django-
cacheops django-cacheops] invalidated_update() also doesn't work. Seems
it's bug of this extension.
--
Ticket URL: <https://code.djangoproject.com/ticket/28676#comment:3>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
--
You received this message because you are subscribed to the Google Groups
"Django updates" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-updates/065.c6469564dcd553f2ef9c6248f2dbb697%40djangoproject.com.
For more options, visit https://groups.google.com/d/optout.
