open-consul/website/source/intro/getting-started/services.html.markdown

layout	page_title	sidebar_current	description
intro	Registering Services	gettingstarted-services	A service can be registered either by providing a service definition or by making the appropriate calls to the HTTP API. A configuration file is the most common, so we will use this approach to register a service, and then query that service using the REST API and DNS interfaces.

Registering Services

In the previous step, we ran our first agent, saw the cluster members (well, our cluster member), and queried that node. In this guide, we'll register our first service and query that service.

Defining a Service

A service can be registered either by providing a service definition or by making the appropriate calls to the HTTP API.

A service definition is the most common way to register services, so we'll use that approach for this step. We'll be building on the agent configuration we covered in the previous step.

First, create a directory for Consul configuration. Consul loads all configuration files in the configuration directory, so a common convention on Unix systems is to name the directory something like /etc/consul.d (the .d suffix implies "this directory contains a set of configuration files").

$ sudo mkdir /etc/consul.d

Next, we'll write a service definition configuration file. Let's pretend we have a service named "web" running on port 80. Additionally, we'll give it a tag we can use as an additional way to query the service:

$ echo '{"service": {"name": "web", "tags": ["rails"], "port": 80}}' \
    | sudo tee /etc/consul.d/web.json

Now, restart the agent, providing the configuration directory:

$ consul agent -dev -config-dir=/etc/consul.d
==> Starting Consul agent...
...
    [INFO] agent: Synced service 'web'
...

You'll notice in the output that it "synced" the web service. This means that the agent loaded the service definition from the configuration file, and has successfully registered it in the service catalog.

If you wanted to register multiple services, you could create multiple service definition files in the Consul configuration directory.

Querying Services

Once the agent is started and the service is synced, we can query the service using either the DNS or HTTP API.

DNS API

Let's first query our service using the DNS API. For the DNS API, the DNS name for services is NAME.service.consul. By default, all DNS names are always in the consul namespace, though this is configurable. The service subdomain tells Consul we're querying services, and the NAME is the name of the service.

For the web service we registered, these conventions and settings yield a fully-qualified domain name of web.service.consul:

$ dig @127.0.0.1 -p 8600 web.service.consul
...

;; QUESTION SECTION:
;web.service.consul.		IN	A

;; ANSWER SECTION:
web.service.consul.	0	IN	A	172.20.20.11

As you can see, an A record was returned with the IP address of the node on which the service is available. A records can only hold IP addresses.

You can also use the DNS API to retrieve the entire address/port pair as a SRV record:

$ dig @127.0.0.1 -p 8600 web.service.consul SRV
...

;; QUESTION SECTION:
;web.service.consul.		IN	SRV

;; ANSWER SECTION:
web.service.consul.	0	IN	SRV	1 1 80 Armons-MacBook-Air.node.dc1.consul.

;; ADDITIONAL SECTION:
Armons-MacBook-Air.node.dc1.consul. 0 IN A	172.20.20.11

The SRV record says that the web service is running on port 80 and exists on the node Armons-MacBook-Air.node.dc1.consul.. An additional section is returned by the DNS with the A record for that node.

Finally, we can also use the DNS API to filter services by tags. The format for tag-based service queries is TAG.NAME.service.consul. In the example below, we ask Consul for all web services with the "rails" tag. We get a successful response since we registered our service with that tag:

$ dig @127.0.0.1 -p 8600 rails.web.service.consul
...

;; QUESTION SECTION:
;rails.web.service.consul.		IN	A

;; ANSWER SECTION:
rails.web.service.consul.	0	IN	A	172.20.20.11

HTTP API

In addition to the DNS API, the HTTP API can be used to query services:

$ curl http://localhost:8500/v1/catalog/service/web
[{"Node":"Armons-MacBook-Air","Address":"172.20.20.11","ServiceID":"web", \
	"ServiceName":"web","ServiceTags":["rails"],"ServicePort":80}]

The catalog API gives all nodes hosting a given service. As we will see later with health checks you'll typically want to query just for healthy instances where the checks are passing. This is what DNS is doing under the hood. Here's a query to look for only healthy instances:

$ curl 'http://localhost:8500/v1/health/service/web?passing'
[{"Node":"Armons-MacBook-Air","Address":"172.20.20.11","Service":{ \
	"ID":"web", "Service":"web", "Tags":["rails"],"Port":80}, "Checks": ...}]

Updating Services

Service definitions can be updated by changing configuration files and sending a SIGHUP to the agent. This lets you update services without any downtime or unavailability to service queries.

Alternatively, the HTTP API can be used to add, remove, and modify services dynamically.

Next Steps

We've now configured a single agent and registered a service. This is good progress, but let's explore the full value of Consul by setting up our first cluster!