Deploying multiple apps with different versions using a single Ansible command

Problem description

In this article, we will look at how to deal with the following scenario. Our system consists of several applications. Let’s imagine we need to deploy some (or all) of those apps with a single command. We also need to specify a common version for all apps or different versions for each app. The desired command could look like this:

ansible-playbook -i prod app-1.yml app-2.yml app-3.yml -e app_1_version=1.2.3 -e app_version=3.2.1

Such a command should result in the installation of app1 in version
1.2.3 and remaining apps in version 3.2.1.

The main difficulty here is that we use the same tasks (e.g. common role) for deploying all apps, so the app version variable is dynamic. Additionally, our app names can contain dashes (which is an invalid character in variable names) so we need to replace it with an underscore.

How to get a dynamic variable

First things first. We need to have access to all variables if the names of those variables are dynamic. Variables must be passed in a form that allows us to retrieve their value using a dynamic variable name. This is where a vars variable comes to the rescue.

vars is a special variable in the form of a dictionary, where we can get a specific value using the syntax of vars[key]. Now, if we have our app name in a variable app_name we can get that variable using the syntax vars[app_name | replace('-', '_') + '_version']. Alternative syntax for this is lookup('vars', app_name | replace('-', '_') + '_version'), but the first one is more pleasant to us. Anyway, not bad, is it?

Default app version

Here comes our additional requirement – a default app version. That’s
a piece of cake – we just have to check if a variable is defined and
take the default version, right? Both solutions above come with some
default support.

vars[spring_app_name | replace('-', '_') + '_version'] | default(app_version) }}
lookup('vars', spring_app_name | replace('-', '_') + '_version', default = app_version)

All we need to do now is to register it in a fact (like
target_app_version) and we are good to go. This is not very
readable though, with all those braces and filters. Can we do it
better?

Custom filter plugin

We can create our custom filter plugin. Let’s see the usage first and focus on the implementation afterwards. In order to get the app version using a custom filter, we can use
something like this:

app_name | ver(vars)

To create a filter, we must add it in the folder filter_plugins of our role. Filters are written in python, so let’s add the following content to ROLE_NAME/filter_plugins/ver.py:

class FilterModule(object):
    def filters(self):
        return {
            'ver': self.ver
        }

    @staticmethod
    def ver(app_name, vars):
        app_name_version = app_name.replace('-', '_') + '_version'
        if app_name_version in vars:
            return vars[app_name_version]
        elif 'app_version' in vars:
            return vars['app_version']
        else:
            raise Exception('No ' + app_name_version + ' or app_version defined')

This solution also has an additional advantage – it raises a meaningful and descriptive error, which is much better than in previous solutions.

Summary

What we achieved is the ability to retrieve the app version from the command line parameter even if multiple apps were deployed at once. Additionally, we are able to specify the default version if a specific version for the app is not defined. Finally, we simplified playbooks code by moving complicated formulas to a custom plugin, providing a descriptive error as well.

You May Also Like

Recently at storm-users

I've been reading through storm-users Google Group recently. This resolution was heavily inspired by Adam Kawa's post "Football zero, Apache Pig hero". Since I've encountered a lot of insightful and very interesting information I've decided to describe some of those in this post.

  • nimbus will work in HA mode - There's a pull request open for it already... but some recent work (distributing topology files via Bittorrent) will greatly simplify the implementation. Once the Bittorrent work is done we'll look at reworking the HA pull request. (storm’s pull request)

  • pig on storm - Pig on Trident would be a cool and welcome project. Join and groupBy have very clear semantics there, as those concepts exist directly in Trident. The extensions needed to Pig are the concept of incremental, persistent state across batches (mirroring those concepts in Trident). You can read a complete proposal.

  • implementing topologies in pure python with petrel looks like this:

class Bolt(storm.BasicBolt):
    def initialize(self, conf, context):
       ''' This method executed only once '''
        storm.log('initializing bolt')

    def process(self, tup):
       ''' This method executed every time a new tuple arrived '''       
       msg = tup.values[0]
       storm.log('Got tuple %s' %msg)

if __name__ == "__main__":
    Bolt().run()
  • Fliptop is happy with storm - see their presentation here

  • topology metrics in 0.9.0: The new metrics feature allows you to collect arbitrarily custom metrics over fixed windows. Those metrics are exported to a metrics stream that you can consume by implementing IMetricsConsumer and configure with Config.java#L473. Use TopologyContext#registerMetric to register new metrics.

  • storm vs flume - some users' point of view: I use Storm and Flume and find that they are better at different things - it really depends on your use case as to which one is better suited. First and foremost, they were originally designed to do different things: Flume is a reliable service for collecting, aggregating, and moving large amounts of data from source to destination (e.g. log data from many web servers to HDFS). Storm is more for real-time computation (e.g. streaming analytics) where you analyse data in flight and don't necessarily land it anywhere. Having said that, Storm is also fault-tolerant and can write to external data stores (e.g. HBase) and you can do real-time computation in Flume (using interceptors)

That's all for this day - however, I'll keep on reading through storm-users, so watch this space for more info on storm development.

I've been reading through storm-users Google Group recently. This resolution was heavily inspired by Adam Kawa's post "Football zero, Apache Pig hero". Since I've encountered a lot of insightful and very interesting information I've decided to describe some of those in this post.

  • nimbus will work in HA mode - There's a pull request open for it already... but some recent work (distributing topology files via Bittorrent) will greatly simplify the implementation. Once the Bittorrent work is done we'll look at reworking the HA pull request. (storm’s pull request)

  • pig on storm - Pig on Trident would be a cool and welcome project. Join and groupBy have very clear semantics there, as those concepts exist directly in Trident. The extensions needed to Pig are the concept of incremental, persistent state across batches (mirroring those concepts in Trident). You can read a complete proposal.

  • implementing topologies in pure python with petrel looks like this:

class Bolt(storm.BasicBolt):
    def initialize(self, conf, context):
       ''' This method executed only once '''
        storm.log('initializing bolt')

    def process(self, tup):
       ''' This method executed every time a new tuple arrived '''       
       msg = tup.values[0]
       storm.log('Got tuple %s' %msg)

if __name__ == "__main__":
    Bolt().run()
  • Fliptop is happy with storm - see their presentation here

  • topology metrics in 0.9.0: The new metrics feature allows you to collect arbitrarily custom metrics over fixed windows. Those metrics are exported to a metrics stream that you can consume by implementing IMetricsConsumer and configure with Config.java#L473. Use TopologyContext#registerMetric to register new metrics.

  • storm vs flume - some users' point of view: I use Storm and Flume and find that they are better at different things - it really depends on your use case as to which one is better suited. First and foremost, they were originally designed to do different things: Flume is a reliable service for collecting, aggregating, and moving large amounts of data from source to destination (e.g. log data from many web servers to HDFS). Storm is more for real-time computation (e.g. streaming analytics) where you analyse data in flight and don't necessarily land it anywhere. Having said that, Storm is also fault-tolerant and can write to external data stores (e.g. HBase) and you can do real-time computation in Flume (using interceptors)

That's all for this day - however, I'll keep on reading through storm-users, so watch this space for more info on storm development.