api-breaker
#
Summary#
NameThe plugin implements API fuse functionality to help us protect our upstream business services.
About the breaker timeout logic
the code logic automatically triggers the unhealthy state incrementation of the number of operations.
Whenever the upstream service returns a status code from the unhealthy.http_statuses
configuration (e.g., 500), up to unhealthy.failures
(e.g., three times) and considers the upstream service to be in an unhealthy state.
The first time unhealthy status is triggered, broken for 2 seconds.
Then, the request is forwarded to the upstream service again after 2 seconds, and if the unhealthy.http_statuses
status code is returned, and the count reaches unhealthy.failures
again, broken for 4 seconds.
and so on, 2, 4, 8, 16, 32, 64, ..., 256, 300. 300
is the maximum value of max_breaker_sec
, allow users to specify.
In an unhealthy state, when a request is forwarded to an upstream service and the status code in the healthy.http_statuses
configuration is returned (e.g., 200) that healthy.successes
is reached (e.g., three times), and the upstream service is considered healthy again.
#
AttributesName | Type | Requirement | Default | Valid | Description |
---|---|---|---|---|---|
break_response_code | integer | required | [200, ..., 599] | Return error code when unhealthy | |
max_breaker_sec | integer | optional | 300 | >=3 | Maximum breaker time(seconds) |
unhealthy.http_statuses | array[integer] | optional | {500} | [500, ..., 599] | Status codes when unhealthy |
unhealthy.failures | integer | optional | 3 | >=1 | Number of consecutive error requests that triggered an unhealthy state |
healthy.http_statuses | array[integer] | optional | {200} | [200, ..., 499] | Status codes when healthy |
healthy.successes | integer | optional | 3 | >=1 | Number of consecutive normal requests that trigger health status |
#
How To EnableHere's an example, enable the api-breaker
plugin on the specified route.
Response 500 or 503 three times in a row to trigger a unhealthy. Response 200 once in a row to restore healthy.
curl "http://127.0.0.1:9080/apisix/admin/routes/1" -H 'X-API-KEY: edd1c9f034335f136f87ad84b625c8f1' -X PUT -d '{ "plugins": { "api-breaker": { "break_response_code": 502, "unhealthy": { "http_statuses": [500, 503], "failures": 3 }, "healthy": { "http_statuses": [200], "successes": 1 } } }, "uri": "/hello", "host": "127.0.0.1",}'
#
Test PluginThen. Like the configuration above, if your upstream service returns 500. 3 times in a row. The client will receive a 502 (break_response_code) response.
$ curl -i -X POST "http://127.0.0.1:9080/get"HTTP/1.1 502 Bad GatewayContent-Type: application/octet-streamConnection: keep-aliveServer: APISIX/1.5
... ...
#
Disable PluginWhen you want to disable the api-breader
plugin, it is very simple, you can delete the corresponding json configuration in the plugin configuration, no need to restart the service, it will take effect immediately:
curl http://127.0.0.1:9080/apisix/admin/routes/1 -H 'X-API-KEY: edd1c9f034335f136f87ad84b625c8f1' -X PUT -d '{ "uri": "/hello", "upstream": { "type": "roundrobin", "nodes": { "127.0.0.1:1980": 1 } }}'
The api-breaker
plugin has been disabled now. It works for other plugins.