Puppet Class: role::analytics_test_cluster::coordinator

Defined in:
modules/role/manifests/analytics_test_cluster/coordinator.pp

Overview

Class role::analytics_test_cluster::coordinator

This role includes Oozie and Hive servers, as well as a MySQL instance that stores meta data associated with those and other Analytics Cluster services.

This role is a launch pad for various cron based Hadoop jobs. Many ingestion jobs need a starting point. Oozie is a great Hadoop job scheduler, but it is not better than cron for some jobs that need to be launched at regular time intervals. Cron is used for those. These crons do not use local resources, instead, they launch Hadoop jobs that run throughout the cluster.

This roles sets up a node responsible to coordinate and orchestrate a Hadoop cluster equipped with tools like Camus, Hive, Oozie and the Analytics Refinery.



19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
 
# File 'modules/role/manifests/analytics_test_cluster/coordinator.pp', line 19

class role::analytics_test_cluster::coordinator {

    system::role { 'analytics_test_cluster::coordinator':
        description => 'Analytics Cluster host running various Hadoop services (Hive, Camus, Oozie, ..) and maintenance scripts'
    }

    include ::profile::analytics::cluster::gitconfig

    include ::profile::java
    include ::profile::analytics::cluster::client
    include ::profile::analytics::database::meta

    # This is a Hadoop client, and should
    # have any special analytics system users on it
    # for interacting with HDFS.
    include ::profile::analytics::cluster::users

    # Back up analytics-meta MySQL instance
    # to an-master1002.
    include ::profile::analytics::database::meta::backup

    # SQL-like queries to data stored in HDFS
    include ::profile::hive::metastore
    include ::profile::hive::server
    include ::profile::hive::metastore::database

    # (Faster) SQL-like queries to data stored in HDFS and elsewhere
    # coordinator only runs the Presto server as a coordinator process.
    # The actual workers are configured in the presto::server role.
    # This node is marked as a coordinator in hiera.
    include ::profile::presto::server

    # The Hadoop job scheduler
    include ::profile::oozie::server
    include ::profile::oozie::server::database

    # Include a weekly cron job to run hdfs balancer.
    include ::profile::hadoop::balancer

    # kafkatee + kafkacat set up to read only a small
    # subset of webrequest traffic and send it to a testing
    # topic.
    include ::profile::kafkatee::webrequest::analytics

    # Various crons that launch Hadoop jobs.
    include ::profile::analytics::refinery

    # Camus crons import data into
    # from Kafka into HDFS.
    include ::profile::analytics::refinery::job::test::camus
    #include ::profile::analytics::refinery::job::test::data_purge
    include ::profile::analytics::refinery::job::test::druid_load
    include ::profile::analytics::refinery::job::test::refine

    include ::profile::kerberos::keytabs

    include ::profile::standard
    include ::profile::base::firewall

    include ::profile::kerberos::client

    # Temporary rule to test JupyterHub + YarnSpawner.
    # Notebook Serviers running in Yarn Hadoop Workers=
    # need to be able to contact JupyterHub.
    # Bug: T224658
    ferm::service{ 'jupyterhub_hub':
        proto  => 'tcp',
        port   => '8081',
        srange => '$ANALYTICS_NETWORKS',
    }
}