Question: lua-resty-upstream with lua-resty-redis

[Blasterdick]

Can this be used with lua-resty-redis to connect to an upstream of redis servers? If yes, provide me some example, if you please. Thanks in advance.

[hamishforbes]

Yes, you simply pass your 'resty.redis' instance to the connect method.
As in the example on the connect docs https://github.com/hamishforbes/lua-resty-upstream#connect

You must configure your pools and hosts first as per the overview example

[Blasterdick]

Thank you, I'll try to do as you said.

But for now, I'm facing an error while configuring openresty with this module:

./configure: error: no /var/chef/cache/lua-resty-upstream/config was found

I just cloned it with git and passed the path to the directory to a configure parameters as --add-module=/var/chef/cache/lua-resty-upstream.

What am I doing wrong? Thanks in advance.

[hamishforbes]

This is a pure Lua module, it doesn't need compiling into openresty.

You simply put it somewhere that your lua_package_path will find it

[Blasterdick]

Great thank you, my bad.

Now, I'm using a config like next:

nginx.conf:

lua_package_path "/var/www/lua-resty-upstream/lib/?.lua;/usr/share/lua/5.1/?.lua;/usr/local/share/lua/5.1/?.lua;;";

lua_shared_dict upstream_dict 1m;
init_by_lua_file /var/www/app/upstream.lua;

init_worker_by_lua 'upstream:init_background_thread()';

server {
}

upstream.lua:

upstream_socket  = require("resty.upstream.socket")
upstream_api = require("resty.upstream.api")

upstream, configured = upstream_socket:new("upstream_dict")
if not upstream then
    error(configured)
end
api = upstream_api:new(upstream)

if not configured then -- Only reconfigure on start, shared mem persists across a HUP
    api:create_pool({id = "primary", timeout = 5})
    api:set_priority("primary", 0)
    api:set_method("primary", "round_robin")
    api:add_host("primary", { id="a", host = "127.0.0.1", port = "6379", weight = 10 })
    api:add_host("primary", { id="b", host = "127.0.0.1", port = "6380", weight = 10 })
    api:add_host("primary", { id="c", host = "127.0.0.1", port = "6381", weight = 10 })
    api:add_host("primary", { id="d", host = "127.0.0.1", port = "6382", weight = 10 })
end

app.lua:

resty_redis = require('resty.redis')
local redis = resty_redis.new()

redis:set_timeout(5000)

local key = ngx.req.get_headers()["X-Forwarded-For"]

local redis, err = upstream:connect(redis, key)

if not redis then
    ngx.log(ngx.ERR, err)
    ngx.status = 500
    return ngx.exit(ngx.status)
end

ngx.log(ngx.info, 'Connected to ' .. err.host.host .. ':' .. err.host.port)
if string.len(post_args) > 0 then

    local res, err = redis:rpush(ngx.var.redis_key, data)

    if not res then
        ngx.say("RPUSH: failed to read reply: ", err)
        return
    end

else
    -- ERROR
    ngx.exit(ngx.HTTP_INTERNAL_SERVER_ERROR)
end

But I'm seeing in nginx log an error like next:

2017/08/18 13:26:00 [error] 26312#0: *4327 attempt to send data on a closed socket: u:0000000000000000, c:0000000000000000, ft:0 eof:0, client: aaa.bbb.ccc.ddd.eee, server: _, request: "POST /app?bla=blabla HTTP/1.1", host: "my.domain.com"

What am I doing wrong? Thanks in advance.

[hamishforbes]

You're setting the timeout on the pool to 5ms, that is probably too low...
api:create_pool({id = "primary", timeout = 5})

If that doesn't fix it can you set your log level to debug and post the full log?

I'd also suggest wrapping your code up in a module rather than creating global vars
See https://github.com/openresty/lua-nginx-module#data-sharing-within-an-nginx-worker for details

[Blasterdick]

Thanks for a suggestion, man.

I set it to 1000, and the error is the same.

How can I set debug level of the Lua logs (I'm a newbie in Lua, and google shows only error_log format recommendations)?
Thanks in advance.

[hamishforbes]

I mean the nginx error_log log level :)