kbrebanov / ansible-nomad

Ansible nomad role

nomad

Installs and configures Nomad

Requirements

This role requires Ansible 1.9 or higher.

Role Variables

Name	Default	Description
nomad_version	0.4.1	Version of Nomad to install
nomad_sha256sum	0cdb5dd95c918c6237dddeafe2e9d2049558fea79ed43eacdfcd247d5b093d67	SHA 256 checksum of package
nomad_client	true	Install Nomad client (nomad_server must be false)
nomad_server	false	Install Nomad server (nomad_client must be false)
nomad_telemetry	false	Enable Nomad telemetry
nomad_atlas	false	Enable Atlas
nomad_region	"global"	Specifies the region the Nomad agent is a member of
nomad_datacenter	"dc1"	Datacenter of the local agent
nomad_name	''	The name of the local node
nomad_log_level	"INFO"	Controls the verbosity of logs the Nomad agent will output. Valid log levels include "WARN", "INFO", or "DEBUG" in increasing order of verbosity
nomad_bind_addr	"127.0.0.1"	Used to indicate which address the Nomad agent should bind to for network services, including the HTTP interface as well as the internal gossip protocol and RPC mechanism
nomad_enable_debug	false	Enables the debugging HTTP endpoints
nomad_ports_http	4646	The port used to run the HTTP server
nomad_ports_rpc	4647	The port used for internal RPC communication between agents and servers, and for inter-server traffic for the consensus algorithm (raft)
nomad_ports_serf	4648	The port used for the gossip protocol for cluster membership
nomad_addresses_http	"{{ nomad_bind_addr }}"	The address the HTTP server is bound to
nomad_addresses_rpc	"{{ nomad_bind_addr }}"	The address to bind the internal RPC interfaces to
nomad_addresses_serf	"{{ nomad_bind_addr }}"	The address used to bind the gossip layer to
nomad_advertise_rpc	"{{ nomad_bind_addr }}"	The address to advertise for the RPC interface
nomad_advertise_serf	"{{ nomad_bind_addr }}"	The address advertised for the gossip layer
nomad_leave_on_interrupt	false	Enables gracefully leaving when receiving the interrupt signal
nomad_leave_on_terminate	false	Enables gracefully leaving when receiving the terminate signal
nomad_enable_syslog	false	Enables logging to syslog
nomad_syslog_facility	"LOCAL0"	Controls the syslog facility that is used
nomad_disable_update_check	false	Disables automatic checking for security bulletins and new version releases
nomad_disable_anonymous_signature	false	Disables providing an anonymous signature for de-duplication with the update check
nomad_telemetry_statsite_address	''	Address of a statsite server to forward metrics data to
nomad_telemetry_statsd_address	''	Address of a statsd server to forward metrics to
nomad_telemetry_disable_hostname	false	Indicates if gauge values should not be prefixed with the local hostname
nomad_server_bootstrap_expect	3	This is an integer representing the number of server nodes to wait for before bootstrapping
nomad_server_data_dir	"{{ nomad_data_dir }}/server"	This is the data directory used for server-specific data, including the replicated log
nomad_server_protocol_version	''	The Nomad protocol version spoken when communicating with other Nomad servers
nomad_server_num_schedulers	"{{ ansible_processor_cores }}"	The number of parallel scheduler threads to run. This can be as many as one per core, or 0 to disallow this server from making any scheduling decisions
nomad_server_enabled_schedulers	[]	This is an array of strings indicating which sub-schedulers this server will handle
nomad_client_state_dir	"{{ nomad_data_dir }}/client"	This is the state dir used to store client state
nomad_client_alloc_dir	"{{ nomad_data_dir }}/alloc"	A directory used to store allocation data
nomad_client_servers	[]	An array of server addresses
nomad_client_node_id	''	This is the value used to uniquely identify the local agent's node registration with the servers. This can be any arbitrary string but must be unique to the cluster. By default, if not specified, a randomly- generate UUID will be used
nomad_client_node_class	''	A string used to logically group client nodes by class. This can be used during job placement as a filter
nomad_client_meta	{}	This is a key/value mapping of metadata pairs
nomad_atlas_infrastructure	''	The Atlas infrastructure name to connect this agent to
nomad_atlas_token	''	The Atlas token to use for authentication
nomad_atlas_join	false	Indicates if the auto-join feature of Atlas should be enabled
nomad_atlas_endpoint	''	The address of the Atlas instance to connect to
nomad_server_node_gc_threshold	''	Controls how long a node must be in a terminal state before it is garbage collected and purged from the system
nomad_server_rejoin_after_leave	''	When provided, Nomad will ignore a previous leave and attempt to rejoin the cluster when starting
nomad_server_retry_join	[]	Similar to start_join but allows retrying a join if the first attempt fails
nomad_server_retry_interval	30s	The time to wait between join attempts
nomad_server_retry_max	0	The maximum number of join attempts to be made before exiting with a return code of 1. By default, this is set to 0 which is interpreted as infinite retries
nomad_server_start_join	[]	An array of strings specifying addresses of nodes to join upon startup. If Nomad is unable to join with any of the specified addresses, agent startup will fail.

Dependencies

• kbrebanov.unzip

Example Playbook

Install Nomad

- hosts: all
  roles:
    - kbrebanov.nomad

License

BSD

Author Information

Kevin Brebanov

ansible-nomad's Issues

nomad_advertise_serf isn't respected

It's documented, but only nomad_addresses_serf is used in nomad.hcl.

When run with nomad_server set, script in /etc/init/nomad.conf is blank

I used this playbook:

- hosts: nomad_servers
  vars:
    use_nomad: true
    nomad_server: true
    nomad_server_bootstrap_expect: "{{ play_hosts|length }}"
  roles:
    - role: kbrebanov.nomad
      tags: nomad
      when: use_nomad
      sudo: yes

and the result was:

ENDOR\marca@sjc-cmdb01:~$ cat /etc/init/nomad.conf
description "Service for nomad"
author "Kevin Brebanov"

start on filesystem or runlevel [2345]
stop on shutdown

reload signal SIGHUP

respawn
respawn limit 10 10
kill timeout 10

script
end script

post-stop script
    rm -f /var/run/nomad.pid
end script

Note how the script for launching nomad is blank.

Nomad download & unzip fails

I'm currently evaluating nomad and this role looked the best to get the job done fast - thank you for creating it 👍

I've run into an issue though - our machines never access 'root' user and we have a special user which can do certain things which are permitted via sudo.
Given that:

• for some reason I can't override nomad_download_dir to /tmp
• when doing it by hand I'm running into permission errors because become: true is set in the download and unzip tasks

Is there any reason for using root user here? Feels like always downloading to tmp or ~ of the current user would do the trick.

Happy to send a PR which addresses this issue, but it's always better to check first ;-) - I might be missign something.

Thanks