Skip to main content

Perfmetrics configuration for Zope/Plone.

Project description

Perfmetrics configuration for Zope and Plone

zperfmetrics works like and depends on perfmetrics, additional it:

  • offers a ZConfig configuration for the statsd,

  • uses the current requests path in the statd dotted path prefix,

  • also provides a patch to measure plone.app.theming for diazo compile and diazo transform if Plone is available,

Installation

First get your statsd configured properly. This is out of scope of this module.

Depend on this module in your packages setup.py.

Given you use zc.builodut to set up your Zope2 or Plone, add to your [instance] section or the section with recipe = plone.recipe.zope2instance the lines:

[instance01]
recipe = plone.recipe.zope2instance
...
zope-conf-imports =
    zperfmetrics
zope-conf-additional =
    <perfmetrics>
        uri statsd://localhost:8125
        before MyFancyProject
        hostname on
        virtualhost on
        after ${:_buildout_section_name_}
    </perfmetrics>
...

Given this runs on a host called w-plone1, this will result in a prefix MyFancyProject.w-plone1.instance01.

uri

Full URI of statd.

before

Prefix path before the hostname.

hostname

Get hostname and insert into prefix. (Boolean: on or off)

virtualhost

Get virtualhost and use in ZPerfmetrics after the static prefix. (Boolean: on or off)

after

Prefix path after the hostname.

Usage

You need to decorate your code or use the with statement to measure a code block.

Usage:

from zperfmetrics import ZMetric
from zperfmetrics import zmetric
from zperfmetrics import zmetricmethod

@zmetric
def some_function():
    # whole functions timing and call counts will be measured
    pass

@ZMetric(stat='a_custom_prefix')
def some_other_function():
    # set a custom prefix instead of module path and fucntion name.
    pass

class Foo(object):

    @zmetricmethod
    def some_method(self):
        # whole methods timing and call counts will be measured
        pass

    @ZMetric(method=True, timing=False):
    def some_counted_method(self):
        # whole methods number of calls will be measured, but no times
        pass

    @ZMetric(method=True, count=False):
    def some_timed_method(self):
        # whole methods timing be measured, but no counts
        pass

    def some_method_partly_measured(self):
        # do something here you dont want to measure
        with ZMetric(stat='part_of_other_method_time'):
            # do something to measure
            # call counts and timing will be measured
            pass
        # do something here you dont want to measure

    @ZMetric(method=True, rate=0.5):
    def some_random_recorded_method(self):
        # randomly record 50% of the calls.
        pass

Request Lifecycle Integration

All ZPerfmetrics with a request passed in are considered to be under the request_lifecycle section.

All metrics in here are build like: $PREFIX.request_lifecycle.[$VIRTUAL_HOST].$PATH.*.

Zope

This package provides subscribers to measure the time a request takes, including some points in time between.

These subscribers are loaded via zcml and are logging under publish.*:

publish.traversal

time needed from publication start until traversal is finished.

publish.rendering

time needed from traversal end until before commit begin.

This value is a bit fuzzy and should be taken with a grain of salt, because there can be other subscribers to this event which take their time. Since the order of execution of the subscribers is not defined, processing may happen after this measurement

If plone tranformchain is active, the rendering time is before transforms are starting.

publish.finalize

time needed from rendering end (or transform end if plone.transformchain is active) until database commit is done.

publish_all

whole time needed from publication start until request is completly processed.

Plone

Installing this package in Plone by depending on zperfmetrics[plone] forces usage of plone.transformchain version 1.2 or newer.

First, publish.rendering gets less fuzzy because the expensive transforms (also subscribers to publish.beforecommit) are all done.

Then it introduces new measurements related to plone.transformchain:

publish.transform_all

time needed for all transforms in the plone.transformchain. This usually includes Diazo.

publish_transform_single.${ORDER}-${TRANSFORMNAME}

time needed for a specific single transform. transforms are ordered and named, both are replaced.

This package patches:

diazo.setup metric

plone.app.theming.transform.ThemeTransform.setupTransform is patched as a basic (path-less) perfmetrics Metric. The setup of the transform happens once on startup and is the time needed to create the Diazo xslt from its rules.xml, index.html and related files.

Statd, Graphite & Grafana in Docker

Setting up Statsd, Graphite and Grafana can be complex. For local testing - but also for production environments - firing up some docker containers comes in handy.

A very minimal version of such a Statd, Graphite & Grafana in Docker setup (original) helps getting things initially up and running. Install Docker and install docker-compose (just pip install docker-compose), then just clone the repository and in its docker directory run docker-compose up -d.

Let Zperfmetrics point to uri statsd://localhost:8125 and collect some data. Open Grafana in your browser at http://localhost:3000.

Go first to Grafana Getting Started, the 10 minute video really helps to find the hidden part of its UI.

Source Code

The sources are in a GIT DVCS with its main branches at github.

We’d be happy to see many branches, forks and pull-requests to make zperfmetrics even better.

Contributors

History

1.0b5 (2016-08-05)

  • Event better naming in order to be able to group and stack: Renamed publish.sum to publish_all and publish.transform_single to publish_transform_single. [jensens]

1.0b4 (2016-08-05)

  • Cleanup names, better grouping in order to make graphing easier. [jensens]

1.0b3 (2016-08-05)

  • Enhancement: Introduce after-prefix a top level request_lifecycle to be consistent with levels. [jensens]

1.0b2 (2016-08-04)

  • Fix: Name of a single transform may contain dots. They are now replaced by _. [jensens]

  • Fix: The root element of a path was not shown on the same level as all other paths. Now it is shown as __root__ [jensens]

  • Feature: It was not possible to filter by virtual host in environment with more than one site per Zope. A new configuration option virtualhost was introduced. If set to on, the virtualhost name will be inserted before PATH. [jensens]

  • Support: In order to make a test installation easier, an example docker-compose setup was added to the Github repository together with some documentation in the README.rst. [jensens]

1.0b1 (2016-08-04)

  • Added subscribers to plone.transformchain events. New setup extra [plone]. Removed one patch for diazo transform. Made publish.beforecommit less fuzzy in Plone. [jensens]

1.0a7 (2016-08-03)

  • Fix: virtual path config in ZMetric class. [syzn]

1.0a6 (2016-07-28)

  • Feature: New value publish.commit. Delta time used from publish.beforecommit to request end. [jensens]

  • Fix: publish.sum shows now overall time of request processing. [jensens]

  • Fix: Update README to reflect last changes. [jensens]

  • Use more beautiful paths if VirtualHostMonster is used. [syzn]

1.0a5 (2016-06-28)

  • Fix: Before/after hooks were not assigned correctly. [jensens]

1.0a4 (2016-06-10)

  • Fixes measurements of zope request, also add more detailed metrics. [jensens]

1.0a3 (2016-06-09)

  • Measure time a zope request needs. [jensens]

1.0a2 (2016-03-22)

  • Refactored: ZConfig best practice [jensens]

  • Fix: Strip trailing dot from prefix [jensens]

1.0a1 (2016-03-22)

  • Fix: README was wrong. [jensens]

1.0a0 (2016-03-22)

  • made it work [jensens, 2016-03-17]

License

Copyright (c) 2016, BlueDynamics Alliance, Austria, Germany, Switzerland All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

  • Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

  • Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

  • Neither the name of the BlueDynamics Alliance nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY BlueDynamics Alliance AS IS AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL BlueDynamics Alliance BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

zperfmetrics-1.0b5.tar.gz (13.3 kB view details)

Uploaded Source

File details

Details for the file zperfmetrics-1.0b5.tar.gz.

File metadata

File hashes

Hashes for zperfmetrics-1.0b5.tar.gz
Algorithm Hash digest
SHA256 bf0b201f5bc39db9de6fad88f256082c2c4e17c54ec38d29e52d79512c75e888
MD5 216e4be35ba97dc854e35d32c18fbf39
BLAKE2b-256 c2f61a838c694e90a3608327ac0dd14d6a894e04c26b2d550b6266185d08f712

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page