Open Stack

On Mon, Aug 1, 2016 at 2:29 PM, Kevin Benton <kevin at benton.pub> wrote:

> >We still want the exception to rollback the entire API operation and
> stopping it with a nested operation I think would mess that up.
>
> Well I think you would want to start a nested transaction, capture the
> duplicate, call the ipam delete methods, then throw a retryrequest. The
> exception will still trigger a rollback of the entire operation.
>

This is kind of what I was headed when I decided to solicit some feedback.
It is a possibility should still be considered.


> >Second, I've been throwing around the idea of not sharing the session
> with the IPAM driver.
>
> If the IPAM driver does not have access to the session, it can't see any
> of the uncommitted data. Would that be a problem? In particular, doesn't
> the IPAM driver's DB table have foreign key constraints with the data
> waiting to be committed in the other session? I'm hesitant to take this
> approach because it means other (if the in-tree doesn't already) IPAM
> drivers cannot have any relational integrity with the objects in question.
>

The in-tree driver doesn't have any FK constraints back to the neutron db
schema for IPAM [1]. I don't think that would make sense since it is
supposed to work like an external driver.


> A related question is, why does the in-tree IPAM driver have to do
> anything at all on a rollback? It currently does share a session which is
> automatically going to rollback all of it's DB operations for it. If it's
> because the driver cannot distinguish a delete call from a rollback and a
> normal delete, I suggest we change the delete call to pass a flag
> indicating that it's for a rollback. That would allow any DB-based drivers
> to just do nothing at this step.
>

Given that it shares the session, it wouldn't have to do anything. But,
again, it wouldn't behave like an external driver. I'd like to not have
special drivers that behave differently than drivers that are really
external; we end up finding things that the in-tree driver does in our
testing that doesn't work right for other drivers.

Drivers might need to access uncommitted data from the neutron DB. I think
even external drivers do this. However, there is a hard line between the
Neutron tables (even IPAM related ones) and the pluggable IPAM driver
database schema. I should have been a little more explicit that I wasn't
suggesting that we hide the Neutron session from the driver. What I meant
to suggest is that we use a different session for the part of the database
schema that belongs solely to the driver. All of the changes would be
inside the driver implementation and the interface to the driver wouldn't
change at all.

Carl

[1]
https://github.com/openstack/neutron/blob/2b1c143ca9/neutron/ipam/drivers/neutrondb_ipam/db_models.py
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160801/084d8ce5/attachment.html>