Heron supports deployment on Apache Mesos. Heron can also run on Mesos using Apache Aurora as a scheduler or using a local scheduler.
Heron's Mesos scheduler interacts with Mesos to stand up all of the components necessary to manage topologies.
To run Heron on Mesos, you'll need to set up a ZooKeeper cluster and configure Heron to communicate with it. Instructions can be found in Setting up ZooKeeper.
In order to deploy Heron, your Mesos cluster will need to have access to a variety of Heron binaries, which can be hosted wherever you‘d like, so long as it’s accessible to Mesos (for example in Amazon S3 or using a local blog storage solution). You can build those binaries using the instructions in Creating a New Heron Release.
Once your Heron binaries are hosted somewhere that's accessible to Mesos, you should run tests to ensure that Mesos can successfully fetch them.
Note: Setting up a Heron cluster involves changing a series of configuration files in the actual heron repository itself (for example ZooKeeper configuration). You should build Heron releases for deployment only after you've made the necessary configuration changes.
Once you‘ve set up a way to host Heron release binaries, you should modify the Mesos configuration in heron/cli2/src/python/mesos_scheduler.conf. In particular, you can specify a URI using the heron.core.release.uri parameter. Here’s an example:
heron.core.release.uri:http://s3.amazonaws.com/my-heron-binaries
Once you've set up ZooKeeper and generated a Mesos-accessible Heron release, any machine that has the heron-cli tool can be used to manage Heron topologies (i.e. can submit topologies, activate and deactivate them, etc.).
The most important thing at this stage is to ensure that heron-cli is synced across all machines that will be working with topologies. Once that has been ensured, you can use Mesos as a scheduler by specifying the proper configuration and configuration loader when managing topologies.
You‘ll need to specify a scheduler configuration at all stages of a topology’s lifecycle by using the --config-file flag to point at a configuration file. There is a default Mesos configuration located in the Heron repository at heron/cli/src/python/mesos_scheduler.conf. You can use this file as is, modify it, or use an entirely different configuration.
Here's an example CLI command using this configuration:
$ heron-cli activate \ # Set scheduler overrides, point to a topology JAR, etc. --config-file=/Users/janedoe/heron/heron/cli/src/python/mesos_scheduler.conf` \ # Other parameters
You can use Heron‘s Mesos configuration loader by setting the --config-loader flag to org.apache.heron.scheduler.aurora.MesosConfigLoader. Here’s an example CLI command:
$ heron-cli submit \ # Set scheduler overrides, point to a topology JAR, etc. --config-loader=org.apache.heron.scheduler.mesos.MesosConfigLoader \ # Other parameters