Malone does not cope well with concurrent updates

[Christian Reis]

I just set the assignee of a bugtask at the same time as Celso was doing it. I got a system error:

ProgrammingError: ERROR: could not serialize access due to concurrent update UPDATE BugTask SET assignee = 247596 WHERE id = 2715

We were changing the same row and same field to the same value. I don't think Malone needs to detect that the changes were identical, but it can't give me a system error.

[Stuart Bishop (stub)]

Christian Reis wrote:
> Public bug reported:
> https://launchpad.net/malone/bugs/2610
>
> Affects: upstream malone
> Severity: Normal
> Priority: Medium
> Status: New
>
> Description:
> I just set the assignee of a bugtask at the same time as Celso was doing it. I
> got a system error:
>
> ProgrammingError: ERROR: could not serialize access due to concurrent update
> UPDATE BugTask SET assignee = 247596 WHERE id = 2715
>
> We were changing the same row and same field to the same value. I don't think
> Malone needs to detect that the changes were identical, but it can't give me a
> system error.

We need to catch this particular error (not just psycopg.ProgrammingError,
but this particular 'serialize access' one) and make Z3 raise the relevant
ConflictError that causes it to retry the transaction. (Dealing with this
case is the resposibility of the application - these exceptions are to be
expected when using the serialize transaction isolation level)

--
Stuart Bishop <email address hidden> http://www.canonical.com/
Canonical Ltd. http://www.ubuntu.com/

[Dafydd Harries (daf)]

Looks like bug #2088 is the one about making our DB adaptor catch concurrency errors.