[openstack-dev] RPC duplicate message bug fix.

Eric Windisch eric at cloudscaling.com
Thu Feb 21 16:59:23 UTC 2013


Bringing this back to the mailing list. We are discussing these reviews:
 https://review.openstack.org/#/c/20567/
 https://review.openstack.org/#/c/22495/
 https://review.openstack.org/#/c/22495/


>
> These lines come from a period in time when I thought Eric did not
> want duplicate message checking to be in ZMQ and was contemplating
> putting the change in common.py and having a check in common code to
> not do it for ZMQ.  I didn't like the idea of checking implementation
> specific stuff in common code, so I was pushing for Kei's change.
> Once Eric cleaned up https://review.openstack.org/22495 it became
> apparent that he was OK with doing duplicate message checking with
> ZMQ, so I simply abstained from the decision.
 
There may have been confusion here, too. I never meant to suggest
checking implementations or doing implementation-specific things in common.

Yes, I'm okay with the duplicate message checking being able to affect
ZMQ, although I still suggest it may be of value to have this be a
toggle as the value is particularly situational. (Generally message
signing *or* mirrored queues - neither are default scenarios)

By the way, I'm bringing *this* discussion back to the list. I'm
admittedly biased, but I do opine that the AMQP-specific solution is
bringing technical debt. I have nothing against the AMQP-specific patch
directly, except that I foresee a need to do this in rpc-common for
Havana, and we'll be making a mess for backwards compatibility.  I know
Mark prefers the AMQP solution, although I'm still not clear why.  I
*do* acknowledge that Mark has specific concerns over minor details of
my patch, which is entirely fine, for which I'd happily make adjustments
and resubmit.

Despite my reservations, I do *not* intend to -1 the AMQP-specific patch
over my own bias, excepting specific details or obvious errors in the
code. If the AMQP-specific patch gets accepted, I will be accepting of
it. I won't do anything as harsh as submitting an unexpected or
undiscussed revert patch.

Finally, Ray and Kei, I thank you both for being very understanding
through the entire discussion over this patch series, both through our
public and our private conversations. It is not common to have such
friction and in no way should either of you be discouraged from
continuing your very useful and valued contributions.

Regards,
Eric Windisch



More information about the OpenStack-dev mailing list