Alex Dadgar 77dbd975f8 resource usage

2016-06-29 14:31:39 -07:00

24 KiB

Raw Blame History

layout	page_title	sidebar_current	description
docs	Configuration	docs-agent-config	Learn about the configuration options available for the Nomad agent.

Configuration

Nomad agents have a variety of options that can be specified via configuration files or command-line flags. Config files must be written in HCL or JSON syntax. Nomad can read and combine options from multiple configuration files or directories to configure the Nomad agent.

Loading Configuration Files

When specifying multiple config file options on the command-line, the files are loaded in the order they are specified. For example:

nomad agent -config server.conf -config /etc/nomad -config extra.json

Will load configuration from server.conf, from .hcl and .json files under /etc/nomad, and finally from extra.json.

Configuration files in directories are loaded alphabetically. With the directory option, only files ending with the .hcl or .json extensions are used. Directories are not loaded recursively.

As each file is processed, its contents are merged into the existing configuration. When merging, any non-empty values from the latest config file will append or replace options in the current configuration. An empty value means "" for strings, 0 for integer or float values, and false for booleans. Since empty values are ignored you cannot disable an option like server mode once you've enabled it.

Complex data types like arrays or maps are usually merged. Some configuration options can also be specified using the command-line interface. Please refer to the sections below for the details of each option.

Configuration Syntax

The preferred configuration syntax is HCL, which supports comments, but you can also use JSON. Below is an example configuration file in HCL syntax.

bind_addr = "0.0.0.0"
data_dir = "/var/lib/nomad"

advertise {
  # We need to specify our host's IP because we can't
  # advertise 0.0.0.0 to other nodes in our cluster.
  rpc = "1.2.3.4:4647"
}

server {
  enabled = true
  bootstrap_expect = 3
}

client {
  enabled = true
  network_speed = 10
  options {
    "driver.raw_exec.enable" = "1"
  }
}

consul {
    # Consul's HTTP Address
    address = "1.2.3.4:8500"
}

atlas {
  infrastructure = "hashicorp/mars"
  token = "atlas.v1.AFE84330943"
}

Note that it is strongly recommended not to operate a node as both client and server, although this is supported to simplify development and testing.

General Options

The following configuration options are available to both client and server nodes, unless otherwise specified:

region: Specifies the region the Nomad agent is a member of. A region typically maps to a geographic region, for example us, with potentially multiple zones, which map to datacenters such as us-west and us-east. Defaults to global.
datacenter: Datacenter of the local agent. All members of a datacenter should share a local LAN connection. Defaults to dc1.
name: The name of the local node. This value is used to identify individual nodes in a given datacenter and must be unique per-datacenter. By default this is set to the local host's name.
data_dir: A local directory used to store agent state. Client nodes use this directory by default to store temporary allocation data as well as cluster information. Server nodes use this directory to store cluster state, including the replicated log and snapshot data. This option is required to start the Nomad agent and must be specified as an absolute path.
log_level: Controls the verbosity of logs the Nomad agent will output. Valid log levels include WARN, INFO, or DEBUG in increasing order of verbosity. Defaults to INFO.
bind_addr: 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. This should be specified in IP format, and can be used to easily bind all network services to the same address. It is also possible to bind the individual services to different addresses using the addresses configuration option. Defaults to the local loopback address 127.0.0.1.
enable_debug: Enables the debugging HTTP endpoints. These endpoints can be used with profiling tools to dump diagnostic information about Nomad's internals. It is not recommended to leave this enabled in production environments. Defaults to false.
ports: Controls the network ports used for different services required by the Nomad agent. The value is a key/value mapping of port numbers, and accepts the following keys:
- http: The port used to run the HTTP server. Applies to both client and server nodes. Defaults to 4646.
- rpc: The port used for internal RPC communication between agents and servers, and for inter-server traffic for the consensus algorithm (raft). Defaults to 4647. Only used on server nodes.
- serf: The port used for the gossip protocol for cluster membership. Both TCP and UDP should be routable between the server nodes on this port. Defaults to 4648. Only used on server nodes.
addresses: Controls the bind address for individual network services. Any values configured in this block take precedence over the default bind_addr. The value is a map of IP addresses and supports the following keys:
- http: The address the HTTP server is bound to. This is the most common bind address to change. Applies to both clients and servers.
- rpc: The address to bind the internal RPC interfaces to. Should be exposed only to other cluster members if possible. Used only on server nodes, but must be accessible from all agents.
- serf: The address used to bind the gossip layer to. Both a TCP and UDP listener will be exposed on this address. Should be restricted to only server nodes from the same datacenter if possible. Used only on server nodes.
advertise: Controls the advertise address for individual network services. This can be used to advertise a different address to the peers of a server or a client node to support more complex network configurations such as NAT. This configuration is optional, and defaults to the bind address of the specific network service if it is not provided. The value is a map of IP addresses and ports and supports the following keys:
- http: The address to advertise for the HTTP interface. This should be reachable by all the nodes from which end users are going to use the Nomad CLI tools.
```
advertise {
   http = "1.2.3.4:4646"
}
```
- rpc: The address to advertise for the RPC interface. This address should be reachable by all of the agents in the cluster. For example:
```
advertise {
  rpc = "1.2.3.4:4647"
}
```
- serf: The address advertised for the gossip layer. This address must be reachable from all server nodes. It is not required that clients can reach this address.
consul: The consul configuration block changes how Nomad interacts with Consul. Nomad can automatically advertise Nomad services via Consul, and can automatically bootstrap itself using Consul. For more details see the consul section.

telemetry: Used to control how the Nomad agent exposes telemetry data to external metrics collection servers. This is a key/value mapping and supports the following keys:
- statsite_address: Address of a statsite server to forward metrics data to.
- statsd_address: Address of a statsd server to forward metrics to.
- disable_hostname: A boolean indicating if gauge values should not be prefixed with the local hostname.
leave_on_interrupt: Enables gracefully leaving when receiving the interrupt signal. By default, the agent will exit forcefully on any signal.
leave_on_terminate: Enables gracefully leaving when receiving the terminate signal. By default, the agent will exit forcefully on any signal.
enable_syslog: Enables logging to syslog. This option only works on Unix based systems.
syslog_facility: Controls the syslog facility that is used. By default, LOCAL0 will be used. This should be used with enable_syslog.
disable_update_check: Disables automatic checking for security bulletins and new version releases.
disable_anonymous_signature: Disables providing an anonymous signature for de-duplication with the update check. See disable_update_check.
http_api_response_headers: This object allows adding headers to the HTTP API responses. For example, the following config can be used to enable CORS on the HTTP API endpoints:
```
http_api_response_headers {
    Access-Control-Allow-Origin = "*"
}
```
atlas: See the atlas options for more details.

Consul Options

The following options are used to configure Consul integration and are entirely optional.

consul: The top-level config key used to contain all Consul-related configuration options. The value is a key/value map which supports the following keys:
- address: The address to the local Consul agent given in the format of host:port. Defaults to 127.0.0.1:8500, which is the same as the Consul default HTTP address.
- token: Token is used to provide a per-request ACL token. This options overrides the Consul Agent's default token.
- auth: The auth information to use for http access to the Consul Agent given as username:password.
- ssl: This boolean option sets the transport scheme to talk to the Consul Agent as https. Defaults to false.
- verify_ssl: This option enables SSL verification when the transport scheme for the Consul API client is https. Defaults to true.
- ca_file: Optional path to the CA certificate used for Consul communication, defaults to the system bundle if not specified.
- cert_file: The path to the certificate used for Consul communication. If this is set then you need to also set key_file.
- key_file: The path to the private key used for Consul communication. If this is set then you need to also set cert_file.
- server_service_name: The name of the service that Nomad registers servers with. Defaults to nomad.
- client_service_name: The name of the service that Nomad registers clients with. Defaults to nomad-client.
- auto_advertise: When enabled Nomad advertises its services to Consul. The services are named according to server_service_name and client_service_name. Nomad Servers and Clients advertise their respective services, each tagged appropriately with either http or rpc tag. Nomad Servers also advertise a serf tagged service. Defaults to true.
- server_auto_join: Servers will automatically discover and join other Nomad Servers by searching for the Consul service name defined in the server_service_name option. This search only happens if the Server does not have a leader. Defaults to true.
- client_auto_join: Client will automatically discover Servers in the Client's region by searching for the Consul service name defined in the server_service_name option. The search occurs if the Client is not registered with any Servers or it is unable to heartbeat to the leader of the region, in which case it may be partitioned and searches for other Servers. Defaults to true

When server_auto_join, client_auto_join and auto_advertise are all enabled, which is by default, and Consul is available, the Nomad cluster will self-bootstrap.

Atlas Options

NOTE: Nomad integration with Atlas is awaiting release of Atlas features for Nomad support. Nomad currently only validates configuration options for Atlas but does not use them. See #183 for more details.

The following options are used to configure Atlas integration and are entirely optional.

atlas: The top-level config key used to contain all Atlas-related configuration options. The value is a key/value map which supports the following keys:
- infrastructure: The Atlas infrastructure name to connect this agent to. This value should be of the form <org>/<infrastructure>, and requires a valid token authorized on the infrastructure.
- token: The Atlas token to use for authentication. This token should have access to the provided infrastructure.
- join: A boolean indicating if the auto-join feature of Atlas should be enabled. Defaults to false.
- endpoint: The address of the Atlas instance to connect to. Defaults to the public Atlas endpoint and is only used if both infrastructure and token are provided.

Server-specific Options

The following options are applicable to server agents only and need not be configured on client nodes.

server: This is the top-level key used to define the Nomad server configuration. It is a key/value mapping which supports the following keys:
- enabled: A boolean indicating if server mode should be enabled for the local agent. All other server options depend on this value being set. Defaults to false.
- bootstrap_expect: This is an integer representing the number of server nodes to wait for before bootstrapping. It is most common to use the odd-numbered integers 3 or 5 for this value, depending on the cluster size. A value of 1 does not provide any fault tolerance and is not recommended for production use cases.
- data_dir: This is the data directory used for server-specific data, including the replicated log. By default, this directory lives inside of the data_dir in the "server" sub-path.
- protocol_version: The Nomad protocol version spoken when communicating with other Nomad servers. This value is typically not required as the agent internally knows the latest version, but may be useful in some upgrade scenarios.
- num_schedulers: 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. This defaults to the number of CPU cores.
- enabled_schedulers: This is an array of strings indicating which sub-schedulers this server will handle. This can be used to restrict the evaluations that worker threads will dequeue for processing. This defaults to all available schedulers.
- node_gc_threshold This is a string with a unit suffix, such as "300ms", "1.5h" or "25m". Valid time units are "ns", "us" (or "µs"), "ms", "s", "m", "h". Controls how long a node must be in a terminal state before it is garbage collected and purged from the system.
- rejoin_after_leave When provided, Nomad will ignore a previous leave and attempt to rejoin the cluster when starting. By default, Nomad treats leave as a permanent intent and does not attempt to join the cluster again when starting. This flag allows the previous state to be used to rejoin the cluster.
- retry_join Similar to start_join but allows retrying a join if the first attempt fails. This is useful for cases where we know the address will become available eventually. Use retry_join with an array as a replacement for start_join, do not use both options.
- retry_interval The time to wait between join attempts. Defaults to 30s.
- retry_max 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.
- 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. By default, the agent won't join any nodes when it starts up. Addresses can be given as an IP, a domain name, or an IP:Port pair. If the port isn't specified the default Serf port, 4648, is used. DNS names may also be used.

Client-specific Options

The following options are applicable to client agents only and need not be configured on server nodes.

client: This is the top-level key used to define the Nomad client configuration. Like the server configuration, it is a key/value mapping which supports the following keys:
- enabled: A boolean indicating if client mode is enabled. All other client configuration options depend on this value. Defaults to false.
- state_dir: This is the state dir used to store client state. By default, it lives inside of the data_dir, in the "client" sub-path. It must be specified as an absolute path.
- alloc_dir: A directory used to store allocation data. Depending on the workload, the size of this directory can grow arbitrarily large as it is used to store downloaded artifacts for drivers (QEMU images, JAR files, etc.). It is therefore important to ensure this directory is placed some place on the filesystem with adequate storage capacity. By default, this directory lives under the data_dir at the "alloc" sub-path. It must be specified as an absolute path.
- servers: An array of server addresses. This list is used to register the client with the server nodes and advertise the available resources so that the agent can receive work. If a port is not specified in the array of server addresses, the default port 4647 will be used.
- node_class: A string used to logically group client nodes by class. This can be used during job placement as a filter. This option is not required and has no default.
- meta: This is a key/value mapping of metadata pairs. This is a free-form map and can contain any string values.
- options: This is a key/value mapping of internal configuration for clients, such as for driver configuration. Please see here for a description of available options.
- network_interface: This is a string to force network fingerprinting to use a specific network interface
- network_speed: This is an int that sets the default link speed of network interfaces, in megabits, if their speed can not be determined dynamically.
- max_kill_timeout: max_kill_timeout is a time duration that can be specified using the s, m, and h suffixes, such as 30s. If a job's task specifies a kill_timeout greater than max_kill_timeout, max_kill_timeout is used. This is to prevent a user being able to set an unreasonable timeout. If unset, a default is used.
- reserved: reserved is used to reserve a portion of the nodes resources from being used by Nomad when placing tasks. It can be used to target a certain capacity usage for the node. For example, 20% of the nodes CPU could be reserved to target a CPU utilization of 80%. The block has the following format:
```
reserved {
    cpu = 500
    memory = 512
    disk = 1024
    reserved_ports = "22,80,8500-8600"
}
```
  - cpu: cpu is given as MHz to reserve.
  - memory: memory is given as MB to reserve.
  - disk: disk is given as MB to reserve.
  - reserved_ports: reserved_ports is a comma separated list of ports to reserve on all fingerprinted network devices. Ranges can be specified by using a hyphen separated the two inclusive ends.

Client Options Map

The following is not an exhaustive list of options that can be passed to the Client, but rather the set of options that configure the Client and not the drivers. To find the options supported by an individual driver, see the drivers documentation here

driver.whitelist: A comma separated list of whitelisted drivers (e.g. "docker,qemu"). If specified, drivers not in the whitelist will be disabled. If the whitelist is empty, all drivers are fingerprinted and enabled where applicable.
env.blacklist: Nomad passes the host environment variables to exec, raw_exec and java tasks. env.blacklist is a comma separated list of environment variable keys not to pass to these tasks. If specified, the defaults are overridden. The following are the default:
- CONSUL_TOKEN
- VAULT_TOKEN
- ATLAS_TOKEN
- AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_SESSION_TOKEN
- GOOGLE_APPLICATION_CREDENTIALS
user.blacklist: An operator specifiable blacklist of users which a task is not allowed to run as when using a driver in user.checked_drivers. Defaults to:
- root
- Administrator
user.checked_drivers: An operator specifiable list of drivers to enforce the the user.blacklist. For drivers using containers, this enforcement often doesn't make sense and as such the default is set to:
- exec
- qemu
- java
fingerprint.whitelist: A comma separated list of whitelisted fingerprinters. If specified, fingerprinters not in the whitelist will be disabled. If the whitelist is empty, all fingerprinters are used.

Command-line Options

A subset of the available Nomad agent configuration can optionally be passed in via CLI arguments. The agent command accepts the following arguments:

alloc-dir=<path>: Equivalent to the Client alloc_dir config option.
-atlas=<infrastructure>: Equivalent to the Atlas infrastructure config option.
-atlas-join: Equivalent to the Atlas join config option.
-atlas-token=<token>: Equivalent to the Atlas token config option.
-bind=<address>: Equivalent to the bind_addr config option.
-bootstrap-expect=<num>: Equivalent to the bootstrap_expect config option.
-client: Enable client mode on the local agent.
-config=<path>: Specifies the path to a configuration file or a directory of configuration files to load. Can be specified multiple times.
-data-dir=<path>: Equivalent to the data_dir config option.
-dc=<datacenter>: Equivalent to the datacenter config option.
-dev: Start the agent in development mode. This enables a pre-configured dual-role agent (client + server) which is useful for developing or testing Nomad. No other configuration is required to start the agent in this mode.
-join=<address>: Address of another agent to join upon starting up. This can be specified multiple times to specify multiple agents to join.
-log-level=<level>: Equivalent to the log_level config option.
-meta=<key=value>: Equivalent to the Client meta config option.
-network-interface<interface>: Equivalent to the Client network_interface config option.
-network-speed<MBits>: Equivalent to the Client network_speed config option.
-node=<name>: Equivalent to the name config option.
-node-class=<class>: Equivalent to the Client node_class config option.
-region=<region>: Equivalent to the region config option.
-rejoin: Equivalent to the rejoin_after_leave config option.
-retry-interval: Equivalent to the retry_interval config option.
-retry-join: Similar to -join but allows retrying a join if the first attempt fails.
-retry-max: Similar to the retry_max config option.
-server: Enable server mode on the local agent.
-servers=<host:port>: Equivalent to the Client servers config option.
-state-dir=<path>: Equivalent to the Client state_dir config option.

24 KiB Raw Blame History