needless looping in legacy pipeline code

[gordon chung]

when publishing samples, the pipeline will sort the samples by name and also group them by name... this is legacy code which was previously used so we could effectively filter samples. the location of filtering samples has moved to an earlier stage so the groupby isn't needed anymore. furthermore, the publisher should not be worrying about sorting what it publishes, it should just push stuff out as quick as possible.

removing this should at worse do nothing. at best, it removes additional looping and thus improves performance.

code in question:

    def publish_samples(self, ctxt, samples):
        for meter_name, samples in itertools.groupby(
                sorted(samples, key=operator.attrgetter('name')),
                operator.attrgetter('name')):
            self._publish_samples(0, ctxt, samples)

ref: https://github.com/openstack/ceilometer/blob/master/ceilometer/pipeline.py#L314-L318

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/140772

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/140772
Committed: https://git.openstack.org/cgit/openstack/ceilometer/commit/?id=494689ff9ed52a32208a1c9966d30c50c0a3af3b
Submitter: Jenkins
Branch: master

commit 494689ff9ed52a32208a1c9966d30c50c0a3af3b
Author: gordon chung <email address hidden>
Date: Wed Dec 10 11:53:14 2014 -0500

    remove useless looping in pipeline

    previously, the pipeline filtered at the publisher stage. this
    logic has been moved to an earlier stage yet we still do processing
    at publisher.

    this patch removes legacy code. at worse, it will have no effect on
    performance. at best, we publish quicker as we don't loop and
    groupby needlessly for each batch of samples.

    Change-Id: I2171c067bd8bb293a8478756f8b4938f312614a1
    Closes-Bug: #1401176