Django transaction.commit() needed after Model.objects.filter() when object was saved from another process

Django transaction.commit() needed after Model.objects.filter() when
object was saved from another process

I have two processes running the Django codebase, for various reasons, one
process will update an object as follows:
myObj.aField = "updated"
myObj.save()
And later on the other process attempts to read that object as follows:
def getObj(xxx):
objs = TheModel.objects.filter(xyz=xxx)
for obj in objs:
print obj.aField
When reading the value from the second process, I won't see the updated
value, instead I see the old value. The second time around I run the
function, I do see the change.
I've noticed that if from the second process (the one reading), I change
the function as follows, I get the updated value:
@transaction.commit_manually def getObj(xxx): objs =
TheModel.objects.filter(xyz=xxx) transaction.commit() for obj in objs:
print obj.aField
After adding the decorator @transaction.commit_manually and the
transaction.commit() line right below the filter() call, I do get the
updated value from the field (which was saved from the other process.)
Is there any reason why this is needed? And what is the impact/potential
issues of using transaction.commit() on a function that actually does not
update models at all? I'm not sure why it works, and if it makes sense, so
hopefully someone else ran into this issue.
Thanks,