On Aug 2, 4:09 pm, Brian Tol <[email protected]> wrote:
> We recently ran into an odd situation with some missing data. We're
> using 0.9.1.pre.
>
> Here's what we're seeing:
>
> a. An orderitem got deleted from the order. There were originally 3
> items for the order (as verified by the email receipt, and the
> transaction record in PayPal), but there are now only 2 items in the
> database
>
> b. One item got deleted all together. The customer received an email
> confirmation, and we have a record of the card being processed in
> PayPal. However there is no order in the database.
>
> We didn't receive any error messages, and nothing else seems to be out
> of place. The order associated with the missing orderitem had no admin
> history associated with it: if a user deleted the product, they did so
> via the shell or the dbshell.
>
> Any thoughts as to what happened, or where we could be looking for
> problems? Are there certain conditions where Satchmo decides to delete
> an order?
>
> It seems fairly random at this point: the store gets dozens of orders
> a week.
>
> Thanks for your help in advance!
>
> -Brian

I have also experienced some untraceable strangeness. One of my
clients had a couple of orders that ended up in an odd state. They had
been marked as 'shipped' which was supposed to trigger payment
capture, but this never happened. Nothing was recorded in the Satchmo
log, no "Order payment failures", etc. to indicate what went wrong.

Core devs:
Any reason not to use database transactions to make sure we don't end
up with incomplete/corrupted data? I don't have any evidence that
would have saved my bacon in this particular case. But in the process
of investigating I noticed how many complex views there are creating
several different types of objects in one go but are not wrapped in a
transaction.

The only bit of code I can find that uses transactions currently is
product.forms.ProductImportForm.

In another thread [1] a user identified an issue with the way certain
livesettings are obtained (but this appears to only affect 0.8?).

I'm by no means a db transaction expert, so curious if there are other
known issues that make liberal use of transactions more complicated
than sprinkling commit_on_success decorators throughout the views.

If not, shall I make a ticket and get started on this in my bitbucket
fork?

Cheers,
John-Scott

-- 
You received this message because you are subscribed to the Google Groups 
"Satchmo users" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/satchmo-users?hl=en.

Reply via email to