ngx_redis2 - Nginx upstream module for the Redis 2.0 protocol
This module is not distributed with the Nginx source. See the installation instructions.
This module is already production ready.
This document describes ngx_redis2 v0.15 released on 19 April 2018.
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 $query;
redis2_pass 127.0.0.1:6379;
}
location = /init {
redis2_query del key1;
redis2_query lpush key1 C;
redis2_query lpush key1 B;
redis2_query lpush key1 A;
redis2_pass 127.0.0.1:6379;
}
location = /get {
redis2_query lrange key1 0 -1;
redis2_pass 127.0.0.1:6379;
}This is an Nginx upstream module that makes nginx talk to a Redis 2.x server in a non-blocking way. The full Redis 2.0 unified protocol has been implemented including the Redis pipelining support.
This module returns the raw TCP response from the Redis server. It's recommended to use my lua-redis-parser (written in pure C) to parse these responses into lua data structure when combined with lua-nginx-module.
When used in conjunction with lua-nginx-module, it is recommended to use the lua-resty-redis library instead of this module though, because the former is much more flexible and memory-efficient.
If you only want to use the get redis command, you can try out the HttpRedisModule. It returns the parsed content part of the Redis response because only get is needed to implement.
Another option is to parse the redis responses on your client side yourself.
syntax: redis2_query cmd arg1 arg2 ...
default: no
context: location, location if
Specify a Redis command by specifying its individual arguments (including the Redis command name itself) in a similar way to the redis-cli utility.
Multiple instances of this directive are allowed in a single location and these queries will be pipelined. For example,
location = /pipelined {
redis2_query set hello world;
redis2_query get hello;
redis2_pass 127.0.0.1:$TEST_NGINX_REDIS_PORT;
}then GET /pipelined will yield two successive raw Redis responses
+OK
$5
worldwhile newlines here are actually CR LF (\r\n).
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.
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 $cmds;
redis2_pass 127.0.0.1:6379;
}Note that in the second sample above, the set_unescape_uri directive is provided by the set-misc-nginx-module.
syntax: redis2_literal_raw_query QUERY
default: no
context: location, location if
Specify a raw Redis query but Nginx variables in it will not be not recognized. In other words, you're free to use the dollar sign character ($) in your QUERY argument.
Only One redis command is allowed in the QUERY argument.
syntax: redis2_pass <upstream_name>
syntax: redis2_pass <host>:<port>
default: no
context: location, location if
phase: content
Specify the Redis server backend.
syntax: redis2_connect_timeout <time>
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.
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).
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).
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.
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;
}
}You can use the excellent HttpUpstreamKeepaliveModule with this module to provide 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;
}
}
}Redis provides the select command to switch Redis databaess. This command is no different from other normal commands like get or set. So you can use them in redis2_query directives, for example,
redis2_query select 8;
redis2_query get foo;This module can be served as a non-blocking redis2 client for lua-nginx-module (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 lua-redis-parser 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 .. "]")
';
}Important The nginx variable
$request_bodyonly contains content buffered into memory. If nginx writes the request body to a temporary file (default behavior) the$request_bodywill not include the entire content or may be empty. It's recommended to use$echo_request_bodywhich reads the full request body (regardless of being in buffer or temporary file).
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)
";
}
}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])
endHere we assume that your Redis server is listening on the default port (6379) of the localhost. We also make use of the lua-redis-parser 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 worldA 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.
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 lua-redis-parser library if you're using Lua to access this module's location.
If you want to use the Redis pub/sub feature with this module, then you must note the following limitations:
Redis server returned extra bytes warnings in your nginx's error.log. Such warnings might be rare but just be prepared for it.If you cannot stand these limitations, then you are highly recommended to switch to the lua-resty-redis library for lua-nginx-module.
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.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:
and finally build the source with this module:
wget 'http://nginx.org/download/nginx-1.11.2.tar.gz'
tar -xzvf nginx-1.11.2.tar.gz
cd nginx-1.11.2/
# 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 installStarting from NGINX 1.9.11, you can also compile this module as a dynamic module, by using the --add-dynamic-module=PATH option instead of --add-module=PATH on the
./configure command line above. And then you can explicitly load the module in your nginx.conf via the load_module
directive, for example,
load_module /path/to/modules/ngx_http_redis2_module.so;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:
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.
The openresty-en mailing list is for English speakers.
The openresty mailing list is for Chinese speakers.
Please submit bug reports, wishlists, or patches by
Available on github at openresty/redis2-nginx-module.
redis2_as_json directive to allow emitting JSON directly.Yichun "agentzh" Zhang (章亦春) agentzh@gmail.com, OpenResty Inc.
You'll be very welcomed to submit patches to the author or just ask for a commit bit to the source repository on GitHub.
This module is licenced under the BSD license.
Copyright (C) 2010-2018, by Yichun "agentzh" Zhang (章亦春) agentzh@gmail.com, OpenResty 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.