FAQ

Page Discussion History

HttpRedis2Module

Revision as of 04:19, 25 April 2014 by Agentzh (Talk | contribs)

Contents

Name

ngx_redis2 - Nginx upstream module for the Redis 2.0 protocol

This module is not distributed with the Nginx source. See the installation instructions.

Status

This module is already production ready.

Version

This document describes ngx_redis2 v0.11 released on 24 April 2014.

Synopsis

    location /foo {
        set $value 'first';
        redis2_query set one $value;
        redis2_pass 127.0.0.1:6379;
    }
 
    # GET /get?key=some_key
    location /get {
        set_unescape_uri $key $arg_key;  # this requires ngx_set_misc
        redis2_query get $key;
        redis2_pass foo.com:6379;
    }
 
    # GET /set?key=one&val=first%20value
    location /set {
        set_unescape_uri $key $arg_key;  # this requires ngx_set_misc
        set_unescape_uri $val $arg_val;  # this requires ngx_set_misc
        redis2_query set $key $val;
        redis2_pass foo.com:6379;
    }
 
    # multiple pipelined queries
    location /foo {
        set $value 'first';
        redis2_query set one $value;
        redis2_query get one;
        redis2_query set one two;
        redis2_query get one;
        redis2_pass 127.0.0.1:6379;
    }
 
    location /bar {
        # $ is not special here...
        redis2_literal_raw_query '*1\r\n$4\r\nping\r\n';
        redis2_pass 127.0.0.1:6379;
    }
 
    location /bar {
        # variables can be used below and $ is special
        redis2_raw_query 'get one\r\n';
        redis2_pass 127.0.0.1:6379;
    }
 
    # GET /baz?get%20foo%0d%0a
    location /baz {
        set_unescape_uri $query $query_string; # this requires the ngx_set_misc module
        redis2_raw_query $query;
        redis2_pass 127.0.0.1:6379;
    }
 
    location /init {
        redis2_query del key1;
        redis2_query lpush key1 C;
        redis2_query lpush key1 B;
        redis2_query lpush key1 A;
        redis2_pass 127.0.0.1:6379;
    }
 
    location /get {
        redis2_query lrange key1 0 -1;
        redis2_pass 127.0.0.1:6379;
    }

Description

This is an Nginx upstream module that makes nginx talk to a Redis 2.x server in a non-blocking way. The full Redis 2.0 unified protocol has been implemented including the Redis pipelining support.

This module returns the raw TCP response from the Redis server. It's recommended to use my LuaRedisParser (written in pure C) to parse these responses into lua data structure when combined with HttpLuaModule.

When used in conjunction with HttpLuaModule, it is recommended to use the lua-resty-redis library instead of this module though, because the former is much more flexible and memory-efficient.

If you only want to use the get redis command, you can try out the HttpRedisModule. It returns the parsed content part of the Redis response because only get is needed to implement.

Another option is to parse the redis responses on your client side yourself.

Directives

redis2_query

syntax: redis2_query cmd arg1 arg2 ...

default: no

context: location, location if

Specify a Redis command by specifying its individual arguments (including the Redis command name itself) in a similar way to the redis-cli utility.

Multiple instances of this directive are allowed in a single location and these queries will be pipelined. For example,

    location /pipelined {
        redis2_query set hello world;
        redis2_query get hello;
 
        redis2_pass 127.0.0.1:$TEST_NGINX_REDIS_PORT;
    }

then GET /pipelined will yield two successive raw Redis responses

    +OK
    $5
    world

while newlines here are actually CR LF (\r\n).

redis2_raw_query

syntax: redis2_raw_query QUERY

default: no

context: location, location if

Specify raw Redis queries and nginx variables are recognized in the QUERY argument.

Only one Redis command is allowed in the QUERY argument, or you'll receive an error. If you want to specify multiple pipelined commands in a single query, use the redis2_raw_queries directive instead.

redis2_raw_queries

syntax: redis2_raw_queries N QUERIES

default: no

context: location, location if

Specify N commands in the QUERIES argument. Both the N and QUERIES arguments can take Nginx variables.

Here's some examples

    location /pipelined {
        redis2_raw_queries 3 "flushall\r\nget key1\r\nget key2\r\n";
        redis2_pass 127.0.0.1:6379;
    }
 
    # GET /pipelined2?n=2&cmds=flushall%0D%0Aget%20key%0D%0A
    location /pipelined2 {
        set_unescape_uri $n $arg_n;
        set_unescape_uri $cmds $arg_cmds;
 
        redis2_raw_queries $n $cmds;
 
        redis2_pass 127.0.0.1:6379;
    }

Note that in the second sample above, the set_unescape_uri directive is provided by the HttpSetMiscModule.

redis2_literal_raw_query

syntax: redis2_literal_raw_query QUERY

default: no

context: location, location if

Specify a raw Redis query but Nginx variables in it will not be not recognized. In other words, you're free to use the dollar sign character ($) in your QUERY argument.

Only One redis command is allowed in the QUERY argument.

redis2_pass

syntax: redis2_pass <upstream_name>

syntax: redis2_pass <host>:<port>

default: no

context: location, location if

phase: content

Specify the Redis server backend.

redis2_connect_timeout

syntax: redis2_connect_timeout

default: 60s

context: http, server, location

The timeout for connecting to the Redis server, in seconds by default.

It's wise to always explicitly specify the time unit to avoid confusion. Time units supported are s(seconds), ms(milliseconds), y(years), M(months), w(weeks), d(days), h(hours), and m(minutes).

This time must be less than 597 hours.

redis2_send_timeout

syntax: redis2_send_timeout <time>

default: 60s

context: http, server, location

The timeout for sending TCP requests to the Redis server, in seconds by default.

It's wise to always explicitly specify the time unit to avoid confusion. Time units supported are s(seconds), ms(milliseconds), y(years), M(months), w(weeks), d(days), h(hours), and m(minutes).

redis2_read_timeout

syntax: redis2_read_timeout <time>

default: 60s

context: http, server, location

The timeout for reading TCP responses from the redis server, in seconds by default.

It's wise to always explicitly specify the time unit to avoid confusion. Time units supported are s(seconds), ms(milliseconds), y(years), M(months), w(weeks), d(days), h(hours), and m(minutes).

redis2_buffer_size

syntax: redis2_buffer_size <size>

default: 4k/8k

context: http, server, location

This buffer size is used for reading Redis replies, but it's not required to be as big as the largest possible Redis reply.

This default size is the page size, may be 4k or 8k.

redis2_next_upstream

syntax: redis2_next_upstream [ error | timeout | invalid_response | off ]

default: error timeout

context: http, server, location

Specify which failure conditions should cause the request to be forwarded to another upstream server. Applies only when the value in redis2_pass is an upstream with two or more servers.

Here's an artificial example:

    upstream redis_cluster {
        server 127.0.0.1:6379;
        server 127.0.0.1:6380;
    }
 
    server {
        location /redis {
            redis2_next_upstream error timeout invalid_response;
            redis2_query get foo;
            redis2_pass redis_cluster;
        }
    }

Connection Pool

You can use the excellent HttpUpstreamKeepaliveModule with this module to privide TCP connection pool for Redis.

A sample config snippet looks like this

    http {
        upstream backend {
          server 127.0.0.1:6379;
 
          # a pool with at most 1024 connections
          # and do not distinguish the servers:
          keepalive 1024;
        }
 
        server {
            ...
            location /redis {
                set_unescape_uri $query $arg_query;
                redis2_query $query;
                redis2_pass backend;
            }
        }
    }

Lua Interoperability

This module can be served as a non-blocking redis2 client for HttpLuaModule (but nowadays it is recommended to use the lua-resty-redis library instead, which is much simpler to use and more efficient most of the time). Here's an example using a GET subrequest:

    location /redis {
        internal;
 
        # set_unescape_uri is provided by ngx_set_misc
        set_unescape_uri $query $arg_query;
 
        redis2_raw_query $query;
        redis2_pass 127.0.0.1:6379;
    }
 
    location /main {
        content_by_lua '
            local res = ngx.location.capture("/redis",
                { args = { query = "ping\\r\\n" } }
            )
            ngx.print("[" .. res.body .. "]")
        ';
    }

Then accessing /main yields

    [+PONG\r\n]

where \r\n is CRLF. That is, this module returns the raw TCP responses from the remote redis server. For Lua-based application developers, they may want to utilize the LuaRedisParser library (written in pure C) to parse such raw responses into Lua data structures.

When moving the inlined Lua code into an external .lua file, it's important to use the escape sequence \r\n directly. We used \\r\\n above just because the Lua code itself needs quoting when being put into an Nginx string literal.

You can also use POST/PUT subrequests to transfer the raw Redis request via request body, which does not require URI escaping and unescaping, thus saving some CPU cycles. Here's such an example:

    location /redis {
        internal;
 
        # $echo_request_body is provided by the ngx_echo module
        redis2_raw_query $echo_request_body;
 
        redis2_pass 127.0.0.1:6379;
    }
 
    location /main {
        content_by_lua '
            local res = ngx.location.capture("/redis",
                { method = ngx.HTTP_PUT,
                  body = "ping\\r\\n" }
            )
            ngx.print("[" .. res.body .. "]")
        ';
    }

This yeilds exactly the same output as the previous (GET) sample.

One can also use Lua to pick up a concrete Redis backend based on some complicated hashing rules. For instance,

    upstream redis-a {
        server foo.bar.com:6379;
    }
 
    upstream redis-b {
        server bar.baz.com:6379;
    }
 
    upstream redis-c {
        server blah.blah.org:6379;
    }
 
    server {
        ...
 
        location /redis {
            set_unescape_uri $query $arg_query;
            redis2_query $query;
            redis2_pass $arg_backend;
        }
 
        location /foo {
            content_by_lua "
                -- pick up a server randomly
                local servers = {'redis-a', 'redis-b', 'redis-c'}
                local i = ngx.time() % #servers + 1;
                local srv = servers[i]
 
                local res = ngx.location.capture('/redis',
                    { args = {
                        query = '...',
                        backend = srv
                      }
                    }
                )
                ngx.say(res.body)
            ";
        }
    }

Pipelined Redis Requests by Lua

Here's a complete example demonstrating how to use Lua to issue multiple pipelined Redis requests via this Nginx module.

First of all, we include the following in our nginx.conf file:

    location = /redis2 {
        internal;
 
        redis2_raw_queries $args $echo_request_body;
        redis2_pass 127.0.0.1:6379;
    }
 
    location = /test {
        content_by_lua_file conf/test.lua;
    }

Basically we use URI query args to pass the number of Redis requests and request body to pass the pipelined Redis request string.

And then we create the conf/test.lua file (whose path is relative to the server root of Nginx) to include the following Lua code:

    -- conf/test.lua
    local parser = require "redis.parser"
 
    local reqs = {
        {"set", "foo", "hello world"},
        {"get", "foo"}
    }
 
    local raw_reqs = {}
    for i, req in ipairs(reqs) do
        table.insert(raw_reqs, parser.build_query(req))
    end
 
    local res = ngx.location.capture("/redis2?" .. #reqs,
        { body = table.concat(raw_reqs, "") })
 
    if res.status ~= 200 or not res.body then
        ngx.log(ngx.ERR, "failed to query redis")
        ngx.exit(500)
    end
 
    local replies = parser.parse_replies(res.body, #reqs)
    for i, reply in ipairs(replies) do
        ngx.say(reply[1])
    end

Here we assume that your Redis server is listening on the default port (6379) of the localhost. We also make use of the LuaRedisParser library to construct raw Redis queries for us and also use it to parse the replies.

Accessing the /test location via HTTP clients like curl yields the following output

OK
hello world

A more realistic setting is to use a proper upstream definition for our Redis backend and enable TCP connection pool via the keepalive directive in it.

Redis Publish/Subscribe Support

This module has limited support for Redis publish/subscribe feature. It cannot be fully supported due to the stateless nature of REST and HTTP model.

Consider the following example:

    location /redis {
        redis2_raw_queries 2 "subscribe /foo/bar\r\n";
        redis2_pass 127.0.0.1:6379;
    }

And then publish a message for the key /foo/bar in the redis-cli command line. And then you'll receive two multi-bulk replies from the /redis location.

You can surely parse the replies with the LuaRedisParser library if you're using Lua to access this module's location.

Limitations For Redis Publish/Subscribe

If you want to use the Redis pub/sub feature with this module, then you must note the following limitations:

  • You cannot use HttpUpstreamKeepaliveModule with this Redis upstream. Only short Redis connections will work.
  • There may be some race conditions that produce the harmless Redis server returned extra bytes warnings in your nginx's error.log. Such warnings might be rare but just be prepared for it.
  • You should tune the various timeout settings provided by this module like redis2_connect_timeout and redis2_read_timeout.

If you cannot stand these limitations, then you are highly recommended to switch to the lua-resty-redis library for HttpLuaModule.

Performance Tuning

  • When you're using this module, please ensure you're using a TCP connection pool (provided by HttpUpstreamKeepaliveModule) and Redis pipelining wherever possible. These features will significantly improve performance.
  • Using multiple instance of Redis servers on your multi-core machines also help a lot due to the sequential processing nature of a single Redis server instance.
  • When you're benchmarking performance using something like ab or http_load, please ensure that your error log level is high enough (like warn) to prevent Nginx workers spend too much cycles on flushing the error.log file, which is always non-buffered and blocking and thus very expensive.

Installation

You are recommended to install this module (as well as the Nginx core and many many other goodies) via the ngx_openresty bundle. Check out the installation instructions for setting up ngx_openresty.

Alternatively, you can install this module manually by recompiling the standard Nginx core as follows:

  • Grab the nginx source code from nginx.org, for example, the version 1.2.7 (see nginx compatibility),
  • and then download the latest version of the release tarball of this module from ngx_redis2's file list.
  • and finally build the source with this module:
    wget 'http://nginx.org/download/nginx-1.2.7.tar.gz'
    tar -xzvf nginx-1.2.7.tar.gz
    cd nginx-1.2.7/
 
    # Here we assume you would install you nginx under /opt/nginx/.
    ./configure --prefix=/opt/nginx \
                --add-module=/path/to/redis2-nginx-module
 
    make -j2
    make install

Compatibility

Redis 2.0, 2.2, 2.4, and above should work with this module without any issues. So is the Alchemy Database (aka redisql in its early days).

The following versions of Nginx should work with this module:

  • 1.3.x (last tested: 1.3.7)
  • 1.2.x (last tested: 1.2.7)
  • 1.1.x (last tested: 1.1.5)
  • 1.0.x (last tested: 1.0.10)
  • 0.9.x (last tested: 0.9.4)
  • 0.8.x >= 0.8.31 (last tested: 0.8.54)

Earlier versions of Nginx will not work.

If you find that any particular version of Nginx above 0.8.31 does not work with this module, please consider reporting a bug.

Community

English Mailing List

The openresty-en mailing list is for English speakers.

Chinese Mailing List

The openresty mailing list is for Chinese speakers.

Bugs and Patches

Please submit bug reports, wishlists, or patches by

  1. creating a ticket on the GitHub Issue Tracker,
  2. or posting to the OpenResty community.

Source Repository

Available on github at openresty/redis2-nginx-module.

TODO

  • Add the redis2_as_json directive to allow emitting JSON directly.

Author

Yichun "agentzh" Zhang (章亦春) <agentzh@gmail.com>, CloudFlare Inc.

Getting involved

You'll be very welcomed to submit patches to the author or just ask for a commit bit to the source repository on GitHub.

Copyright & License

This module is licenced under the BSD license.

Copyright (C) 2010-2014, by Yichun "agentzh" Zhang (章亦春) <agentzh@gmail.com>, CloudFlare Inc.

All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

  • Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
  • Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

SEE ALSO