Liveness health check

Overview

Health checks are extremely important in determining the status of an application - in this instance, the Tyk Gateway. Without them, it can be hard to know the actual state of the Gateway.

Depending on your configuration, the Gateway could be using a few components:

The Tyk Dashboard.
RPC
Redis (compulsory).

Any of these components could go down at any given point and it is useful to know if the Gateway is currently usable or not. A good usage of the health check endpoint is for the configuration of a load balancer to multiple instances of the Gateway or as a Kubernetes liveness probe.

The following component status will not be returned:

MongDB or SQL
Tyk Pump

Note

Health check is implemented as per the Health Check Response Format for HTTP APIs RFC

An example of the response from this API is as follows:

{
  "status": "pass",
  "version": "v3.1.1",
  "description": "Tyk GW",
  "details": {
    "redis": {
      "status": "pass",
      "componentType": "datastore",
      "time": "2020-05-19T03:42:55+01:00"
    },
    "dashboard": {
      "status": "pass",
      "componentType": "system",
      "time": "2020-05-19T03:42:55+01:00"
    },
    "rpc": {
      "status": "pass",
      "componentType": "system",
      "time": "2020-05-19T03:42:55+01:00"
    }
  }
}

Status Levels

The following status levels can be returned in the JSON response.

pass: Indicates that all components required for the Gateway to work 100% are available, and there is no impact on your traffic.
warn: Indicates that one of the components is having an outage but your Gateway is able to keep processing traffic. The impact is medium (i.e. no quotas are applied, no analytics, no RPC connection to MDCB).
fail: Indicates that Redis AND the Tyk Dashboard are unavailable, and can and indicate other failures. The impact is high (i.e. no configuration changes are available for API/policies/keys, no quotas are applied, and no analytics).

Configure health check

By default, the liveness health check runs on the /hello path. But it can be configured to run on any path you want to set. For example:

health_check_endpoint_name: "status"

This configures the health check to run on /status instead of /hello.

Refresh Interval

The Health check endpoint will refresh every 10 seconds.

HTTP error code

The Health check endpoint will always return a HTTP 200 OK response if the polled health check endpoint is available on your Tyk Gateway. If HTTP 200 OK is not returned, your Tyk Gateway is in an error state.

For MDCB installations the /hello endpoint can be polled in either your Management or Worker Gateways. It is recommended to use the /hello endpoint behind a load balancer for HA purposes.

Health check examples

The following examples show how the Health check endpoint returns

Open Source installation

The following is returned for a pass status level for the Open Source Gateway:

$ http :8080/hello
HTTP/1.1 200 OK
Content-Length: 156
Content-Type: application/json
Date: Wed, 14 Apr 2021 17:36:09 GMT

{
  "description": "Tyk GW",
  "details": {
    "redis": {
      "componentType": "datastore",
      "status": "pass",
      "time": "2021-04-14T17:36:03Z"
    }
  },
  "status": "pass",
  "version": "v3.1.1"
}

Self Managed

Redis outage

$ http :8080/hello
HTTP/1.1 200 OK
Content-Length: 303
Content-Type: application/json
Date: Wed, 14 Apr 2021 14:58:06 GMT

{
  "description": "Tyk GW",
  "details": {
    "dashboard": {
      "componentType": "system",
      "status": "pass",
      "time": "2021-04-14T14:58:03Z"
    },
    "redis": {
      "componentType": "datastore",
      "output": "storage: Redis is either down or was not configured",
      "status": "fail",
      "time": "2021-04-14T14:58:03Z"
    }
  },
  "status": "warn",
  "version": "v3.1.2"
}

Dashboard outage

$ http :8080/hello
HTTP/1.1 200 OK
Content-Length: 292
Content-Type: application/json
Date: Wed, 14 Apr 2021 15:52:47 GMT

{
  "description": "Tyk GW",
  "details": {
    "dashboard": {
      "componentType": "system",
      "output": "dashboard is down? Heartbeat is failing",
      "status": "fail",
      "time": "2021-04-14T15:52:43Z"
    },
    "redis": {
      "componentType": "datastore",
      "status": "pass",
      "time": "2021-04-14T15:52:43Z"
    }
  },
  "status": "warn",
  "version": "v3.1.2"
}

Dashboard and Redis outage

$ http :8080/hello
HTTP/1.1 200 OK
Content-Length: 354
Content-Type: application/json
Date: Wed, 14 Apr 2021 17:53:33 GMT

{
  "description": "Tyk GW",
  "details": {
    "dashboard": {
      "componentType": "system",
      "output": "dashboard is down? Heartbeat is failing",
      "status": "fail",
      "time": "2021-04-14T17:53:33Z"
    },
    "redis": {
      "componentType": "datastore",
      "output": "storage: Redis is either down or was not configured",
      "status": "fail",
      "time": "2021-04-14T17:53:33Z"
    }
  },
  "status": "fail",
  "version": "v3.1.2"
}

MDCB installation

Management Gateway outage

The Status of the Management Gateway is the same as for a Tyk Self Management installation.

MDCB Worker Gateway RPC outage

$  http :8080/hello
HTTP/1.1 200 OK
Content-Length: 333
Content-Type: application/json
Date: Wed, 14 Apr 2021 17:21:24 GMT

{
  "description": "Tyk GW",
  "details": {
    "redis": {
      "componentType": "datastore",
      "output": "storage: Redis is either down or was not configured",
      "status": "fail",
      "time": "2021-04-14T17:21:16Z"
    },
    "rpc": {
      "componentType": "system",
      "output": "Could not connect to RPC",
      "status": "fail",
      "time": "2021-04-14T17:21:16Z"
    }
  },
  "status": "fail",
  "version": "v3.1.2"
}