A JavaScript implementation of a client for Eureka (https://github.com/Netflix/eureka), the Netflix OSS service registry.
First, install the module into your node project:
npm install eureka-js-client --save
The Eureka module exports a JavaScript function that can be constructed.
import Eureka from 'eureka-js-client';
// Or, if you're not using a transpiler:
const Eureka = require('eureka-js-client').Eureka;
// example configuration
const client = new Eureka({
// application instance information
instance: {
app: 'jqservice',
hostName: 'localhost',
ipAddr: '127.0.0.1',
port: 8080,
vipAddress: 'jq.test.something.com',
dataCenterInfo: {
name: 'MyOwn',
},
},
eureka: {
// eureka server host / port
host: '192.168.99.100',
port: 32768,
},
});
The Eureka client searches for the YAML file eureka-client.yml
in the current working directory. It further searches for environment specific overrides in the environment specific YAML files (e.g. eureka-client-test.yml
). The environment is typically development
or production
, and is determined by environment variables in this order: EUREKA_ENV
, if present, or NODE_ENV
, if present. Otherwise it defaults to development
. The options passed to the constructor overwrite any values that are set in configuration files.
You can configure a custom directory to load the configuration files from by specifying a cwd
option in the object passed to the Eureka
constructor.
const client = new Eureka({
cwd: `${__dirname}/config`,
});
If you wish, you can also overwrite the name of the file that is loaded with the filename
property. You can mix the cwd
and filename
options.
const client = new Eureka({
filename: 'eureka',
cwd: `${__dirname}/config`,
});
client.start();
client.stop();
const instances = client.getInstancesByAppId('YOURSERVICE');
const instances = client.getInstancesByVipAddress('YOURSERVICEVIP');
The client exposes the ability to modify the outgoing request options object prior to a eureka call. This is useful when adding authentication methods such as OAuth, or other custom headers. This will be called on every eureka request, so it highly suggested that any long-lived external calls made in the middleware are cached or memoized. If the middleware returns anything other than an object, the eureka request will immediately fail and perform a retry if configured.
// example using middleware to set-up HTTP authentication
const client = new Eureka({
requestMiddleware: (requestOpts, done) => {
requestOpts.auth = {
user: 'username',
password: 'somepassword'
};
done(requestOpts);
}
});
For AWS environments (dataCenterInfo.name == 'Amazon'
) the client has built-in logic to request the AWS metadata that the Eureka server requires. See Eureka REST schema for more information.
// example configuration for AWS
const client = new Eureka({
// application instance information
instance: {
app: 'jqservice',
port: 8080,
vipAddress: 'jq.test.something.com',
statusPageUrl: 'http://__HOST__:8080/info',
healthCheckUrl: 'http://__HOST__:8077/healthcheck',
homePageUrl: 'http://__HOST__:8080/',
dataCenterInfo: {
name: 'Amazon',
},
},
eureka: {
// eureka server host / port / EC2 region
host: 'eureka.test.mydomain.com',
port: 80,
},
});
Notes:
- Under this configuration, the instance
hostName
andipAddr
will be set to the public host and public IP that the AWS metadata provides. You can seteureka.useLocalMetadata
totrue
to use the private host and private IP address instead. - If you want to register using the IP address as the hostname, set
eureka.preferIpAddress
totrue
. This may be used in combination witheureka.useLocalMetadata
for selecting the private or public IP. - For status and healthcheck URLs, you may use the replacement key of
__HOST__
to use the host from the metadata. - Metadata fetching can be disabled by setting
config.eureka.fetchMetadata
tofalse
if you want to provide your own metadata in AWS environments.
If your have multiple availability zones and your DNS entries set up according to the Wiki article Deploying Eureka Servers in EC2, you'll want to set config.eureka.useDns
to true
and set config.eureka.ec2Region
to the current region (usually this can be pulled into your application via an environment variable, or passed in directly at startup).
This will cause the client to perform a DNS lookup using config.eureka.host
and config.eureka.ec2Region
. The naming convention for the DNS TXT records required for this to function is also described in the Wiki article above. This feature will also work in non-EC2 environments as long as the DNS records conform to the same convention. The results of the DNS resolution are cached in memory and refreshed every 5 minutes by default (set config.eureka.clusterRefreshInterval
to override).
By default, the client will first try to connect to the Eureka server located in the same availability-zone as it's currently in. If availability-zone
is not set in the instance metadata, a random server will be chosen. This also applies when statically configuring the cluster (mapped by zone, see below). To disable this feature, set config.eureka.preferSameZone
to false
, and a random server will be chosen.
While the recommended approach for resolving the Eureka cluster is using DNS (see above), you can also statically configure the list of Eureka servers by zone or just using a simple default list. Make sure to provide the full protocol, host, port, and path to the Eureka REST service (usually /apps/
) when using this approach.
// example configuration for AWS (static map of Eureka cluster by availability-zone)
const client = new Eureka({
instance: {
... // application instance information
},
eureka: {
availabilityZones: {
'us-east-1': ['us-east-1c', 'us-east-1d', 'us-east-1e']
},
serviceUrls: {
'us-east-1c': [
'http://ec2-fake-552-627-568-165.compute-1.amazonaws.com:7001/eureka/apps/', 'http://ec2-fake-368-101-182-134.compute-1.amazonaws.com:7001/eureka/apps/'
],
'us-east-1d': [...],
'us-east-1e': [...]
}
},
});
// example configuration (static list of Eureka cluster servers)
const client = new Eureka({
instance: {
... // application instance information
},
eureka: {
serviceUrls: {
default: [
'http://ec2-fake-552-627-568-165.compute-1.amazonaws.com:7001/eureka/apps/', 'http://ec2-fake-368-101-182-134.compute-1.amazonaws.com:7001/eureka/apps/'
]
}
},
});
option | default value | description |
---|---|---|
requestMiddleware |
noop | Custom middleware function to modify the outgoing request to eureka |
logger |
console logging | logger implementation for the client to use |
shouldUseDelta |
false | Experimental mode to fetch deltas from eureka instead of full registry on update |
eureka.maxRetries |
3 |
Number of times to retry all requests to eureka |
eureka.requestRetryDelay |
500 |
milliseconds to wait between retries. This will be multiplied by the # of failed retries. |
eureka.heartbeatInterval |
30000 |
milliseconds to wait between heartbeats |
eureka.registryFetchInterval |
30000 |
milliseconds to wait between registry fetches |
eureka.registerWithEureka |
true |
enable/disable Eureka registration |
eureka.fetchRegistry |
true |
enable/disable registry fetching |
eureka.filterUpInstances |
true |
enable/disable filtering of instances with status === UP |
eureka.servicePath |
/eureka/apps/ |
path to eureka REST service |
eureka.ssl |
false |
enable SSL communication with Eureka server |
eureka.useDns |
false |
look up Eureka server using DNS, see Looking up Eureka Servers in AWS using DNS |
eureka.preferSameZone |
true |
enable/disable zone affinity when locating a Eureka server |
eureka.clusterRefreshInterval |
300000 |
milliseconds to wait between refreshing cluster hosts (DNS resolution only) |
eureka.fetchMetadata |
true |
fetch AWS metadata when in AWS environment, see Configuring for AWS environments |
eureka.useLocalMetadata |
false |
use local IP and local hostname from metadata when in an AWS environment. |
eureka.preferIpAddress |
false |
use IP address (local or public) as the hostname for registration when in an AWS environment. |
Eureka client is an instance of EventEmitter
and provides the following events for consumption:
event | data provided | description |
---|---|---|
started |
N/A | Fired when eureka client is fully registered and all registries have been updated. |
registered |
N/A | Fired when the eureka client is registered with eureka. |
deregistered |
N/A | Fired when the eureka client is deregistered with eureka. |
heartbeat |
N/A | Fired when the eureka client has successfully renewed it's lease with eureka. |
registryUpdated |
N/A | Fired when the eureka client has successfully update it's registries. |
The library uses request for all service calls, and debugging can be turned on by passing NODE_DEBUG=request
when you start node. This allows you you double-check the URL being called as well as other request properties.
NODE_DEBUG=request node example.js
You can also turn on debugging within the library by setting the log level to debug:
client.logger.level('debug');
Later versions of Eureka require a slightly different JSON POST body on registration. If you are seeing 400 errors on registration it's probably an issue with your configuration and it could be the formatting differences below. The history behind this is unclear and there's a discussion here. The main differences are:
port
is now an object with 2 required fields$
and@enabled
.dataCenterInfo
has an@class
property.
See below for an example:
const client = new Eureka({
// application instance information
instance: {
app: 'jqservice',
hostName: 'localhost',
ipAddr: '127.0.0.1',
port: {
'$': 8080,
'@enabled': true,
},
vipAddress: 'jq.test.something.com',
dataCenterInfo: {
'@class': 'com.netflix.appinfo.InstanceInfo$DefaultDataCenterInfo',
name: 'MyOwn',
},
},
eureka: {
// eureka server host / port
host: '192.168.99.100',
port: 32768,
},
});
If you are planning on connecting to a eureka service in AWS you will need to add the corresponding dataCenterInfo
information:
dataCenterInfo: {
'@class': 'com.netflix.appinfo.AmazonInfo',
name: 'Amazon',
}
This probably means that the Eureka REST service is located on a different path in your environment. The default is http://<EUREKA_HOST>/eureka/apps
, but depending on your setup you may need to set eureka.servicePath
in your configuration to another path. The REST service could be hung under /eureka/apps/
or possibly /apps/
.
If you are using Spring Cloud you'll likely need the following settings:
- Use the newer style of the configuration here or Spring Cloud Eureka will throw a 500 error.
- Set
statusPageUrl
to a valid URL for your application, Spring Cloud seems to require this when the instance information is parsed. - Put single quotes around boolean
@enabled
. Unfortunately, a 500 error regarding parsing seems to occur without that.
Below is an example configuration that should work with Spring Cloud Eureka server:
const client = new Eureka({
instance: {
app: 'jqservice',
hostName: 'localhost',
ipAddr: '127.0.0.1',
statusPageUrl: 'http://localhost:8080/info',
port: {
'$': 8080,
'@enabled': 'true',
},
vipAddress: 'jq.test.something.com',
dataCenterInfo: {
'@class': 'com.netflix.appinfo.InstanceInfo$DefaultDataCenterInfo',
name: 'MyOwn',
},
},
eureka: {
host: '192.168.99.100',
port: 32768,
servicePath: '/eureka/apps/'
},
});
The test for the module are written using mocha and chai. To run the unit tests, you can use the gulp test
task:
gulp test
If you wish to have the tests watch the src/
and test/
directories for changes, you can use the test:watch
gulp task:
gulp test:watch