Mesos Deployment

❗️

This is a legacy Apache Ignite documentation

The new documentation is hosted here: https://ignite.apache.org/docs/latest/

• Overview
• Ignite Mesos Framework
• Running Ignite Mesos Framework
• Configuration

Overview

Apache Ignite Framework supports scheduling and running Apache Ignite nodes in a Mesos cluster.
Apache Mesos is a cluster manager which provides a general runtime environment providing all the essentials used to deploy, run and manage distributed applications. Its resource management and isolation helps in getting the most out of servers.
For information about Apache Mesos please refer to http://mesos.apache.org/.

Ignite Mesos Framework

Deploying an Apache Ignite cluster typically involves downloading the Apache Ignite distribution, changing configuration settings, and starting the nodes. Apache Ignite Mesos Framework consists of Scheduler and Task, and greatly simplifies cluster deployment.

• Scheduler registers itself at Mesos Master on scheduler initialization. Once the registration is successful, the Scheduler will begin processing resource requests from Mesos Master to utilize resources for Apache Ignite nodes. The scheduler will maintain the Ignite cluster at the desired (and available) resource level (CPU, memory, etc.).
• Task launches an Ignite nodes on Mesos slave.

Running Ignite Mesos Framework

Running Ignite Mesos Framework requires Apache Mesos Cluster configured and running. For information on how to set up a Mesos cluster, please refer to: https://docs.mesosphere.com/getting-started/datacenter/install/

🚧

Make sure that master and slave nodes listen on correct IP addresses. Otherwise, there is no guarantee that Mesos Cluster will function correctly.

Run the Framework via Marathon

Currently, the recommended way to run the framework is to run it via Marathon.

1. Install marathon. See https://mesosphere.github.io/marathon/docs/.
2. Download Apache Ignite and upload the libs\optional\ignite-mesos\ignite-mesos-<ignite-version>.jar file to any cloud storage (for example Amazon S3 storage, etc.).
3. Copy the following application definition (in JSON format) and save to marathon.json file. Update any parameters which you would like to change.

{
  "id": "ignition",
  "instances": 1,
  "cpus": 2,
  "mem": 2048,
  "ports": [0],
  "uris": [
    "http://host/ignite-mesos-<ignite-version>.jar"
  ],
  "env": {
    "IGNITE_NODE_COUNT": "4",
    "MESOS_MASTER_URL": "zk://localhost:2181/mesos",
    "IGNITE_RUN_CPU_PER_NODE": "2",
    "IGNITE_MEMORY_PER_NODE": "2048",
    "IGNITE_VERSION": "1.0.5",
    "MESOS_USER" : "userAAAAA",
    "MESOS_ROLE" :  "role1"
  },
  "cmd": "java -jar ignite-mesos-<ignite-version>.jar"
}

A role name must be a valid directory name, so it cannot:
• Be an empty string
• Be . or ..
• Start with -
• Contain any slash, backspace, or whitespace character

If there are no restrictions set on the cluster, the framework will try to occupy all the resources in the Mesos cluster. See the Configuration section below.

4. Send POST request with the application definition to Marathon, using CURL or your preferred method.

curl -X POST -H "Content-type: application/json" --data-binary @marathon.json http://<marathon-ip>:8080/v2/apps/

5. To make sure that Apache Mesos Framework is deployed correctly: open the Marathon UI at http://<marathon-ip>:8080. Make sure that an application with the name ignition exists and its status is Running.

6. Open Mesos console at http://<master-ip>:5050. If everything works fine, then tasks with names like Ignite node N should have the state RUNNING. In this example, N is equal to 4. See example marathon.json file - "IGNITE_NODE_COUNT": "4"

7. Mesos supports retrieving tasks' logs from the browser. To look through Ignite logs, click on Sandbox in the Active Tasks table.

8. Click on stdout to get stdout logs and on stderr to get stderr logs.

Run the Framework via JAR file

1. Download Ignite package and go to the libs\optional\ignite-mesos\ folder.
2. Run the framework using the following command:

java -jar ignite-mesos-<ignite-version>.jar

or

java -jar ignite-mesos-<ignite-version>.jar properties.prop

where properties.prop is a property file. If the file is not provided, the framework will try to occupy all resources in the Mesos cluster. Here is an example property file:

# The number of nodes in the cluster.
IGNITE_NODE_COUNT=1

# Mesos ZooKeeper URL to locate leading master.
MESOS_MASTER_URL=zk://localhost:2181/mesos

# The number of CPU Cores for each Apache Ignite node.
IGNITE_RUN_CPU_PER_NODE=4

# The number of Megabytes of RAM for each Apache Ignite node.
IGNITE_MEMORY_PER_NODE=4096

# The version ignite which will be run on nodes.
IGNITE_VERSION=1.0.5

3. In order to make sure that Apache Mesos Framework is deployed correctly: open the Mesos console at http://<master-ip>:5050. If everything works fine, the tasks with names like Ignite node N should have the state RUNNING. In this example, N is equal to 1. See example properties.prop file - "IGNITE_NODE_COUNT": "1"

4. Mesos supports retrieving tasks' logs from browser. To look through Ignite logs, click on Sandbox.

5. Click on stdout to get stdout logs and on stderr to get stderr logs.

Configuration

All configurations are handled through environment variables (this lends itself well to being easy to configure marathon to run the framework) or the property file. The following configuration parameters can be optionally configured.