Report status for OpenWhisk Microservices for New Relic (and others) Uptime (HTTP) checks
You have a microservice that is deployed as an OpenWhisk HTTP action or even a number of these microservices and you want to establish monitoring of service uptime using New Relic.
In case the service is down, you want to quickly understand if it is a problem with
- the OpenWhisk runtime, which may be unreachable or overloaded
- one of your backend API providers which might be unreachable
- your own service which could be broken (for instance due to a deployment change)
Finally, you know that there are New Relic Synthetics but you do not want to keep repeating the same code for returning a status check in each of your micro services.
helix-status
is a library that allows you to wrap your own actions to get a standardized monitoring response
$ npm install -S @adobe/helix-status
In the entry point of your action, add
const { helixStatus } = require('@adobe/helix-status');
to the top of your file and override the module.exports.main
with:
module.exports.main = helixStatus(main);
All GET /_status_check/healthcheck.json
requests to your service will now respond with an XML response similar to below:
{
"status": "OK",
"version": "1.5.4",
"response_time": 6,
"process": {}
}
You can also specify a list of checks to run by passing second argument to wrap
:
module.exports.main = helixStatus(main, { example: 'http://www.example.com'})
you will then see results like this:
{
"status": "OK",
"version": "1.5.4",
"response_time": 249,
"process": {
"activation": "7ef3047190924313b3047190923313e9"
},
"example": 247
}
It is a good idea to use URLs that are representative of the API endpoints your service is calling in normal operation as checks.
If you are using Probot for instance through Serverless Probot on OpenWhisk, the usage is slightly different:
// import the probot status app
const { probotStatus } = require('@adobe/helix-status');
probot
.withApp(yourApp)
.withApp(probotStatus()) //add a status check app
probotStatus()
accepts the same checks
object that has been described above, so you can pass an array of URL checks.
New Relic Synthetics is a service that is similar to Pingdom. It can be used with helix-status
by creating an API Check script like this:
const assert = require('assert');
// replace the URL with your check URL
$http.get('https://adobeioruntime.net/api/v1/web/helix/helix-services/status@v3/_status_check/heathcheck.json',
// Callback
function (err, response, body) {
assert.equal(response.statusCode, 200, 'Expected a 200 OK response');
const health = JSON.parse(body);
assert.equal(health.status, 'OK', 'Expected an OK health check status');
for (const v in health) {
if (['status', 'process', 'version'].indexOf(v)===-1) {
$util.insights.set(v, parseInt(health[v]));
}
}
for (const h in ['x-openwhisk-activation-id', 'x-request-id', 'x-version']) {
$util.insights.set(h, response.headers[h]);
}
}
);
By default, helix-status
will take a map of URLs and make a GET
request for each URL provided. In some scenarios, you need to provide additional detail to craft the request and helix-status
supports three types of advanced checks for that:
If you need to adjust things like request method (from GET
to POST
) or set request headers (e.g. Accept
), instead of providing a URL as string, you can provide a request options object, according to the request/request
documentation.
Make sure not to forget the uri
.
module.exports.main = wrap(main,
{ example: {
uri: 'http://www.example.com',
method: 'POST',
headers: {
accept: 'application/json'
}
}});
For more advanced use cases, you can provide a function
in your checks. This function will be executed with the params
of your OpenWhisk function when requested with the status check or health check URLs.
Keep in mind that the check function should execute reasonably fast, but it can be async.
module.exports.main = wrap(main,
{ example: function(params) => params.foo === params.bar });
A combination of the two techniques above is the usage of request options with Dynamic Values. This can be used if you need to populate properties of the request object or request headers with values provided in the params
of the execution.
A typical example would be making a call to an API that requires an API key, where the API key would be stored in the default parameters of the OpenWhisk action.
To achieve this, provide a request options object as described in Request Options, but note that values both of the options
and the options.headers
object can be function
s. Each property that is a function will be replaced with the value that function returns when called with the action's parameters.
module.exports.main = wrap(main,
{ example: {
uri: 'http://www.example.com',
method: 'POST',
headers: {
accept: 'application/json',
authorization: params => `Bearer ${params.EXAMPLE_API_TOKEN}`
}
}});
The example above shows how to extract the EXAMPLE_API_TOKEN
value from the action's (default) parameters and applies it to the Authorization
header.
The health check reports following HTTP status codes:
200
(OK): all health checks performed successfully504
(Gateway Timeout): the health check took too long to execute502
(Gateway Error): the health check got an error response from the checked URL500
(Server Error): the generic check function did not execute successfully
Deploying Helix Status requires the wsk
command line client, authenticated to a namespace of your choice. For Project Helix, we use the helix
namespace.
All commits to main that pass the testing will be deployed automatically. All commits to branches that will pass the testing will get committed as /helix-services/status@ci<num>
and tagged with the CI build number.