Forever subscribe attempts of deepsix

[Yingxin]

When subscribing dataindex/tables from services, the deepsix.schedule() will send 'sub' messages periodically to make sure the subscriptions are successful.

However, the attempts to subscribe will continue forever even after they succeed. So dsebus will be filled with 'sub' messages if there are lots of services and policy rules. Unnecessary routing works of d6cage will introduce additional workloads.

[Rui Chen (kiwik-chenrui)]

Append patch links:

    https://review.openstack.org/#/c/202922/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/202922
Committed: https://git.openstack.org/cgit/openstack/congress/commit/?id=89bd1ab03a39dc40d01668b73061515a53038c7a
Submitter: Jenkins
Branch: master

commit 89bd1ab03a39dc40d01668b73061515a53038c7a
Author: Yingxin Cheng <email address hidden>
Date: Fri Jul 17 16:18:52 2015 +0800

    Stop subscribe attempts after successfully handled

    The 'sub' message will be handled by 'insub()' in publish service. If
    `msg.correlationId` is already in target service's
    subscriberCorrelationUuids list, additional message with the same corrId
    will be simply ignored. So forever subscribe attempts are unnecessary.

    Every successful subscription is replied by a 'pub' message with the
    same correlationId of the corresponding 'sub' message. So if the
    subscriber receives a 'pub' message, it can stop the related subscribe
    attempts by deleting the id from scheduuids.

    Change-Id: Ifd0c68c74c8589a594337fc942088488e2d53fff
    Closes-Bug: #1475531