A proxy service that converts http requests to NATS messages.
This proxy service is designed to be ran in a docker container and will translate all of your http based api calls (REST endpoints). It will translate the http route to a NATS message and will send the message as a request into the NATS messaging system. Once a reply has been returned it is written back to the http response. The primary use case for this service is when you have an ecosystem of microservices that communicate via NATS but you want http REST api clients to be able to communicate to those services as a typical http REST api client.
When an http request is received the http-nats-proxy will take the request path and replace every forward slash ('/')
with a period ('.'). It will then prefix that with the http verb that was used (get, put, post, delete, head, etc.).
Finally it will convert the entire string to lowercase. This becomes your NATS subject where the message will be sent.
For example, the http request of GET http://localhost:5000/test/v1/customer?id=1234&show-history=true will result in a
NATS message subject of get.test.v1.customer
When an http request is received, the http-nats-proxy will extract the headers, cookies, query string parameters and the body of the request and create a NATS message. The NATS message that goes into the messaging system will be a json serialized object that has the following structure:
{
"cookies": {
"key1": "value1",
"key2": "value2"
},
"errorMessage", "", <--- can be set by microservices and pipeline steps
"extendedProperties": { <--- can be set by microservices and pipeline steps
"key1": "value1",
"key2": "value2"
},
"queryParams": {
"key1": "value1",
"key2": "value2"
},
"requestBody": "",
"requestHeaders": {
"key1": "value1",
"key2": "value2"
},
"responseBody": "", <--- can be set by microservices and pipeline steps
"responseHeaders": { <--- can be set by microservices and pipeline steps
"key1": "value1",
"key2": "value2"
},
"responseStatusCode": -1, <--- can be set by microservices and pipeline steps
"shouldTerminateRequest": true|false <--- can be set by microservices and pipeline steps
}
Every http(s) request that comes into the http-nats-proxy can go through a series of steps before being delivered to the
final destination (your microservice). Those steps are defined in a yaml based configuration file and the location of
the file is passed to the https-nats-proxy via the HTTP_NATS_PROXY_REQUEST_PIPELINE_CONFIG_FILE environment variable.
Each step defined in the pipeline configuration file contains the following properties:
subject: The NATS subject name where the request will be delivered for this particular step in the pipeline. In
order to notify the http-nats-proxy to send the message to your microservice use * for the subject name.
pattern: Either publish or request, this tells the http-nats-proxy whether it should do a fire and forget
message exchange pattern or if it should use the request/reply pattern. If you are not going to change or cancel the
http request then you should use publish. If your pipeline step will potentially modify the request or cancel the
request you should use request
direction: Either incoming, outgoing or both. This tells the http-nats-proxy when to call your pipeline step.
incoming is when the message is inbound, outgoing is after the message has reached the end of the pipeline and the
response is being returned. both will call you pipeline step twice, once when the request is inbound and a second time
when the response is outbound.
order: A numeric value that is the order that your step should be called in relation to other steps in the pipeline
There are times when you want to get a copy of the request/response message after it has completed running through all of the pipeline steps. Examples of this would be when you wanted to log the request/response or capture metrics about how long each request took to process, etc. In these cases, the metadata about the request is not available during the execution of the pipeline steps. For these scenarios you can use Observers. Observers are notified via a NATS publish message after all of the pipeline steps have executed and metadata has been stored for the request/response pair. It is a "copy" of the final state of the http request.
steps:
- subject: trace.header
pattern: request
direction: incoming
order: 1
- subject: authentication
pattern: request
direction: incoming
order: 2
- subject: '*'
pattern: request
direction: incoming
order: 3
observers:
- subject: 'pipeline.logging'
- subject: 'pipeline.metrics'
All configuration of the http-nats-proxy is done via environment variables.
| Environment Variable | Default Value | Description |
|---|---|---|
| HTTP_NATS_PROXY_HOST_PORT | 5000 | The port that the http-nats-proxy will listen for incoming http requests on |
| HTTP_NATS_PROXY_NAT_URL | nats://localhost:4222 | The NATS url where the http-nats-proxy will send the NATS message to |
| HTTP_NATS_PROXY_WAIT_TIMEOUT_SECONDS | 10 | The number of seconds that the http-nats-proxy will wait for a response from the microservice backend before it returns a Timeout Error to the http client |
| HTTP_NATS_PROXY_HEAD_STATUS_CODE | 200 | The http status code that will be used for a successful HEAD request |
| HTTP_NATS_PROXY_PUT_STATUS_CODE | 201 | The http status code that will be used for a successful PUT request |
| HTTP_NATS_PROXY_GET_STATUS_CODE | 200 | The http status code that will be used for a successful GET request |
| HTTP_NATS_PROXY_PATCH_STATUS_CODE | 201 | The http status code that will be used for a successful PATCH request |
| HTTP_NATS_PROXY_POST_STATUS_CODE | 201 | The http status code that will be used for a successful POST request |
| HTTP_NATS_PROXY_DELETE_STATUS_CODE | 204 | The http status code that will be used for a successful DELETE request |
| HTTP_NATS_PROXY_CONTENT_TYPE | application/json; charset=utf-8 | The http response Content-Type header value. This should be set to whatever messaging format your microservice api supports (xml, json, etc) |
| HTTP_NATS_PROXY_REQUEST_PIPELINE_CONFIG_FILE | The full file path and name of the configuration file that specifies your request pipeline | |
| LOGGING_LEVEL | Verbose | Verbose Debug Information Warning Error Fatal |
| LOGGING_PROPERTIES | A comma separated list of name/value pairs that will be set as properties on the Serilog log messages |
In order to see the http-nats-proxy in action along with a test microservice, logging and metrics you can run the docker compose file in your environment.
cd src
docker-compose up
This will run the NATS server in a container.
Next start the project from Visual Studio and make sure that each project in the solution is set to run on startup ( multi-project start-up configuration). The http-nats-proxy will listen for http requests on port 5000 of your host machine. You can then send http requests to the proxy and have them processed by the test microservice. The test microservice that comes with this repo will respond to the following http routes:
GET http://localhost:5000/healthcheck
GET http://localhost:5000/customers?id=<any integer>
Additionally, if configured for metrics, logging and tracing you will see the output of those microservices as well in your terminal. This is for demo purposes only. There is a 1/2 second delay in the logging and metrics microservices just to simulate work.
The http-nats-proxy has a Dockerfile where you can build your own docker images by running:
cd src
docker build -t http-nats-proxy .
In order to control what gets returned from the http-nats-proxy, your microservice has to set the response property of
the NATS message that you receive when you subscribe to a NATS subject. You should return the entire NATS message. You
may optionally set the responseStatusCode and the errorMessage properties if an error occurs while you are
processing the message. The nats-http-proxy will honor the responseStatusCode if it is set and will also format and
return an error response if the errorMessage property has been set.