nginx提供健康检查功能，用于检测后端服务不可用时是否转发。

nginx开源版本提供被动健康检查，即应用调用时来进行判断后端服务是否可用，时效性差点；

nginx商业版提供主动健康检查，可以通过nginx内置模块来监控后端服务，若发现不可用时则从组内踢出，相对于被动更加智能灵活。

HTTP Health Checks

Monitor the health of HTTP servers in an upstream group by sending periodic health checks, including customizable active health checks in NGINX Plus.

Introduction

NGINX and NGINX Plus can continually test your upstream servers, avoid the servers that have failed, and gracefully add the recovered servers into the load‑balanced group.

Prerequisites

For passive health checks, NGINX Open Source or NGINX Plus
For active health checks and the live activity monitoring dashboard, NGINX Plus
A load‑balanced group of HTTP upstream servers

Passive Health Checks

For passive health checks, NGINX and NGINX Plus monitor transactions as they happen, and try to resume failed connections. If the transaction still cannot be resumed, NGINX Open Source and NGINX Plus mark the server as unavailable and temporarily stop sending requests to it until it is marked active again.

The conditions under which an upstream server is marked unavailable are defined for each upstream server with parameters to the server directive in the upstream block:

fail_timeout – Sets the time during which a number of failed attempts must happen for the server to be marked unavailable, and also the time for which the server is marked unavailable (default is 10 seconds).
max_fails – Sets the number of failed attempts that must occur during the fail_timeout period for the server to be marked unavailable (default is 1 attempt).

In the following example, if NGINX fails to send a request to a server or does not receive a response from it 3 times in 30 seconds, it marks the server as unavailable for 30 seconds:

upstream backend {
    server backend1.example.com;
    server backend2.example.com max_fails=3 fail_timeout=30s;
}

Note that if there is only a single server in a group, the fail_timeout and max_fails parameters are ignored and the server is never marked unavailable.

Server Slow Start

A recently recovered server can be easily overwhelmed by connections, which may cause the server to be marked as unavailable again. Slow start allows an upstream server to gradually recover its weight from zero to its nominal value after it has been recovered or became available. This can be done with the slow_start parameter of the upstream server directive:

upstream backend {
    server backend1.example.com slow_start=30s;
    server backend2.example.com;
    server 192.0.0.1 backup;
}

Note that if there is only a single server in a group, the slow_start parameter is ignored and the server is never marked unavailable. Slow start is exclusive to NGINX Plus.

Active Health Checks

NGINX Plus can periodically check the health of upstream servers by sending special health‑check requests to each server and verifying the correct response.

To enable active health checks:

In the location that passes requests (proxy_pass) to an upstream group, include the health_check directive:
```
server {
    location / {
        proxy_pass http://backend;
        health_check;
    }
}
```
This snippet defines a server that passes all requests (location /) to the upstream group called backend. It also enables advanced health monitoring with the health_check directive: by default, every five seconds NGINX Plus sends a request for “/” to each server in the backend group. If any communication error or timeout occurs (the server responds with a status code outside the range from 200 through 399) the health check fails. The server is marked as unhealthy, and NGINX Plus does not send client requests to it until it once again passes a health check.

Optionally you can specify another port for health checks, for example, for monitoring health of many services on the same host. Specify a new port with the port parameter of the health_check directive:
```
server {
    location / {
        proxy_pass   http://backend;
        health_check port=8080;
    }
}
```
In the upstream server group, define a shared memory zone with the zone directive:
```
http {
    upstream backend {
        zone backend 64k;
        server backend1.example.com;
        server backend2.example.com;
        server backend3.example.com;
        server backend4.example.com;
    }
}
```
The zone is shared among all worker processes and stores the configuration of the upstream group. This enables the worker processes to use the same set of counters to keep track of responses from the servers in the group.

The defaults for active health checks can be overridden with parameters to the health_check directive:
```
location / {
    proxy_pass   http://backend;
    health_check interval=10 fails=3 passes=2;
}
```
Here, the interval parameter increases the delay between health checks from the default 5 seconds to 10 seconds. The fails parameter requires the server to fail three health checks to be marked as unhealthy (up from the default one). Finally, the passes parameter means the server must pass two consecutive checks to be marked as healthy again instead of the default one.

You can also enable connection caching with the keepalive_time parameter - in case of TLS upstreams the full TLS handshake won’t happen for every health check probe and the connection can be reused during the specified period of time:
```
location / {
    proxy_http_version 1.1;
    proxy_set_header   Connection "";
    proxy_pass         https://backend;
    health_check       interval=1 keepalive_time=60s;
}
```

Specifying the Requested URI

Use the uri parameter of the health_check directive to set the URI to request in a health check:

location / {
    proxy_pass   http://backend;
    health_check uri=/some/path;
}

The specified URI is appended to the server domain name or IP address set for the server in the upstream block. For the first server in the sample backend group declared above, a health check requests the URI http://backend1.example.com/some/path.

Defining Custom Conditions

You can set custom conditions that the response must satisfy for the server to pass the health check. The conditions are defined in a match block, which is referenced in the match parameter of the health_check directive.

On the http {} level, specify the match {} block and name it, for example, server_ok:

http {
    #...
    match server_ok {
        # tests are here
    }
}

Refer to the block from the health_check directive by specifying the match parameter and the name of the match block:

http {
    #...
    match server_ok {
        status 200-399;
        body   !~ "maintenance mode";
    }
    server {
        #...
        location / {
            proxy_pass   http://backend;
            health_check match=server_ok;
        }
    }
}

Here the health check is passed if the status code of the response is in the range 200–399, and its body does not contain the string maintenance mode.

The match directive enables NGINX Plus to check the status code, header fields, and the body of a response. Using this directive it is possible to verify whether the status is in a specified range, whether a response includes a header, or whether the header or body matches a regular expression. The match directive can contain one status condition, one body condition, and multiple header conditions. A response must satisfy all conditions defined in match block for the server to pass the health check.

For example, the following match directive matches responses that have status code 200, the exact value text/html in the Content-Type header, and the text Welcome to nginx! in the body:

match welcome {
    status 200;
    header Content-Type = text/html;
    body   ~ "Welcome to nginx!";
}

The following example uses the exclamation point (!) to define characteristics the response must not have to pass the health check. In this case, the health check passes when the status code is something other than 301, 302, 303, or 307, and there is no Refresh header.

match not_redirect {
    status ! 301-303 307;
    header ! Refresh;
}

Mandatory Health Checks

By default, when a new server is added to an upstream group, NGINX Plus considers it healthy and sends traffic to it immediately. But for some servers, particularly if they were added through the API interface or through DNS resolution, it would be good to perform health check first before allowing them to handle traffic.

The mandatory parameter requires every newly added server to pass all configured health checks before NGINX Plus sends traffic to it.

When combined with slow start, it gives a new server more time to connect to databases and “warm up” before being asked to handle their full share of traffic.

Mandatory health checks can be marked as persistent, so that the previous state is remembered when reloading configuration. Specify the persistent parameter together with the mandatory parameter:

upstream my_upstream {
    zone   my_upstream 64k;
    server backend1.example.com slow_start=30s;
}

server {
    location / {
        proxy_pass   http://my_upstream;
        health_check mandatory persistent;
    }
}

Here the mandatory and persistent parameters of the health_check directive and the slow_start parameter of the server directive are specified. Servers that are added to the upstream group using the API or DNS interfaces are marked as unhealthy and receive no traffic until they pass the health check; at that point they start receiving a gradually increasing amount of traffic over a span of 30 seconds. If NGINX Plus configuration is reloaded and before reload the server was marked as healthy, mandatory health check are not performed and the server state is considered to be up.

Health checks can also be enabled for non-HTTP protocols, such as FastCGI, memcached, SCGI, uwsgi, and also for TCP and UDP.

nginx健康检查