For example, set the default listening port of APISIX to 8000, and keep other configurations as default. The configuration in `conf/config.yaml` should be like this:
Other default configurations can be found in the `conf/config-default.yaml` file, which is bound to the APISIX source code. **Never** manually modify the `conf/config-default.yaml` file. If you need to customize any configuration, you should update the `conf/config.yaml` file.
**Note** `APISIX` will generate `conf/nginx.conf` file automatically, so please *DO NOT EDIT*`conf/nginx.conf` file too.
The route matches the client's request by defining rules, then loads and executes the corresponding plugin based on the matching result, and forwards the request to the specified Upstream.
The route mainly consists of three parts: matching rules (e.g uri, host, remote_addr, etc.), plugin configuration (current-limit & rate-limit, etc.) and upstream information.
The following image shows an example of some Route rules. When some attribute values are the same, the figure is identified by the same color.
We configure all the parameters directly in the Route, it's easy to set up, and each Route has a relatively high degree of freedom. But when our Route has more repetitive configurations (such as enabling the same plugin configuration or upstream information), once we need update these same properties, we have to traverse all the Routes and modify them, so it adding a lot of complexity of management and maintenance.
The shortcomings mentioned above are independently abstracted in APISIX by the two concepts [Service](#service) and [Upstream](#upstream).
A `Service` is an abstraction of an API (which can also be understood as a set of Route abstractions). It usually corresponds to the upstream service abstraction. Between `Route` and `Service`, usually the relationship of N:1, please see the following image.
Different Route rules are bound to a Service at the same time. These Routes will have the same upstream and plugin configuration, reducing redundant configuration.
The following example creates a Service that enables the current-limit plugin, and then binds the Route with the id of `100` and `101` to the Service.
Of course, we can also specify different plugin parameters or upstream for Route. Some of the following Routes have different current-limit parameters. Other parts (such as upstream) continue to use the configuration parameters in Service.
`Plugin` represents the plugin configuration that will be executed during the `HTTP` request/response lifecycle.
The `Plugin` configuration can be bound directly to `Route` or it can be bound to `Service` or `Consumer`. For the configuration of the same plugin, only one copy is valid, and the configuration selection priority is always `Consumer` > `Route` > `Service`.
In `conf/config.yaml`, you can declare which plugins are supported by the local APISIX node. This is a whitelisting mechanism. Plugins that are not in this whitelist will be automatically ignored. This feature can be used to temporarily turn off or turn on specific plugins, which is very effective in dealing with unexpected situations. If you want to add new plugins based on existing plugins, you need to copy the data of `plugins` node which in `conf/config-default.yaml` to the `plugins` node of `conf/config.yaml`.
The configuration of the plugin can be directly bound to the specified Route, or it can be bound to the Service, but the plugin configuration in Route has a higher priority.
A plugin will only be executed once in a single request, even if it is bound to multiple different objects (such as Route or Service).
The plugin configuration is submitted as part of Route or Service and placed under `plugins`. It internally uses the plugin name as the hash's key to hold configuration items for different plugins.
```json
{
...
"plugins": {
"limit-count": {
"count": 2,
"time_window": 60,
"rejected_code": 503,
"key": "remote_addr"
},
"prometheus": {}
}
}
```
Not all plugins have specific configuration items. For example, there is no specific configuration item under `prometheus`. In this case, an empty object identifier can be used.
`Script` represents a script that will be executed during the `HTTP` request/response life cycle.
The `Script` configuration can be directly bound to the `Route`.
`Script` and `Plugin` are mutually exclusive, and `Script` is executed first. This means that after configuring `Script`, the `Plugin` configured on `Route` will not be executed.
In theory, you can write arbitrary Lua code in `Script`, or you can directly call existing plugins to reuse existing code.
`Script` also has the concept of execution phase, supporting `access`, `header_filer`, `body_filter` and `log` phase. The system will automatically execute the code of the corresponding phase in the `Script` script in the corresponding phase.
```json
{
...
"script": "local _M = {} \n function _M.access(api_ctx) \n ngx.log(ngx.INFO,\"hit access phase\") \n end \nreturn _M"
Upstream is a virtual host abstraction that performs load balancing on a given set of service nodes according to configuration rules. Upstream address information can be directly configured to `Route` (or `Service`). When Upstream has duplicates, you need to use "reference" to avoid duplication.
As shown in the image above, by creating an Upstream object and referencing it by ID in `Route`, you can ensure that only the value of an object is maintained.
Upstream configuration can be directly bound to the specified `Route` or it can be bound to `Service`, but the configuration in `Route` has a higher priority. The priority behavior here is very similar to `Plugin`.
In addition to the basic complex equalization algorithm selection, APISIX's Upstream also supports logic for upstream passive health check and retry, see the link below.
A distinguishing feature of APISIX from other API gateways is that it allows users to choose different routers to better match free services, making the best choice between performance and freedom.
Set the route that best suits your business needs in the local configuration `conf/config.yaml`.
*`radixtree_uri`: (Default) only use `uri` as the primary index. Support for full and deep prefix matching based on the `radixtree` engine, see [How to use router-radixtree](router-radixtree.md).
*`Prefix match`: Use `*` at the end to represent the given `uri` as a prefix match. For example, `/foo*` allows matching `/foo/`, `/foo/a` and `/foo/b`.
*`match priority`: first try absolute match, if you can't hit absolute match, try prefix match.
*`Any filter attribute`: Allows you to specify any Nginx built-in variable as a filter, such as URL request parameters, request headers, cookies, and so on.
For the API gateway, it is usually possible to identify a certain type of requester by using a domain name such as a request domain name, a client IP address, etc., and then perform plugin filtering and forward the request to the specified upstream, but sometimes the depth is insufficient.
As shown in the image above, as an API gateway, you should know who the API Consumer is, so you can configure different rules for different API Consumers.
|Field|Required|Description|
|---|----|----|
|username|Yes|Consumer Name.|
|plugins|No|The corresponding plugin configuration of the Consumer, which has the highest priority: Consumer > Route > Service. For specific plugin configurations, refer to the [Plugins](#plugin) section.|
In APISIX, the process of identifying a Consumer is as follows:
3. Get the Plugin or Upstream information bound to the Consumer: Complete the different configurations for different Consumers.
To sum up, Consumer is a consumer of certain types of services and needs to be used in conjunction with the user authentication system.
For example, different consumers request the same API, and the gateway service corresponds to different Plugin or Upstream configurations according to the current request user information.
In addition, you can refer to the [key-auth](./plugins/key-auth.md) authentication authorization plugin call logic to help you further understand the Consumer concept and usage.
How to enable a specific plugin for a Consumer, you can see the following example:
Enable basic debug mode just by setting `apisix.enable_debug = true` in `conf/config.yaml` file.
e.g Using both `limit-conn` and `limit-count` plugins for a `/hello` request, there will have a response header called `Apisix-Plugins: limit-conn, limit-count`.
Enable advanced debug mode by modifying the configuration in `conf/debug.yaml` file. Because there will be a check every second, only the checker reads the `#END` flag, and the file would be considered as closed.
The checker would judge whether the file data changed according to the last modification time of the file. If there has any change, reload it. If there was no change, skip this check. So it's hot reload for enabling or disabling advanced debug mode.
|Key|Optional|Description|Default|
|----|-----|---------|---|
|hook_conf.enable|required|Enable/Disable hook debug trace. Target module function's input arguments or returned value would be printed once this option is enabled.|false|
|hook_conf.name|required|The module list name of hook which has enabled debug trace||
|hook_conf.log_level|required|Logging levels for input arguments & returned value|warn|