# Rewrite lookup via MySQL table

**URL:** <https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766>\
**Category:** NGINX\
**Created:** [August 24, 2010, 2:24am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766 "2010-08-24T02:24:30Z")\
**Posts on this page:** 20\
**Page:** 1

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 2:24am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/1 "2010-08-24T02:24:30Z")

</div>

Any chance I can use a MySQL database table to do a lookup to return the  
url to rewrite to?

Let me explain the situation first…

Magento utilizes rewrite rules heavily but does the majority of the  
rewrite within the Front controller. The problem is that it is VERY  
in-efficient and a store with more than 10,000 products is severely  
delay. But, all information is referenced in a table ‘core\_url\_rewrite’  
so my though it so do the lookup first in Nginx and hand off to Magento  
with the non-seo url.

Any idea/though would be GREATLY appreciated.

Brian Tafoya  
Prema Solutions  
[www.premasolutions.com](http://www.premasolutions.com)

Current environment:

Two dedicated servers, one web, one db, both with ridiculous amounts of  
memory and high speed drives.

nginx  
php-fpm  
php 5.3.3  
apc

Posted at Nginx Forum:

> **[Rewrite lookup via MySQL table](https://forum.nginx.org/read.php?2,122977,122977#msg-122977)**
>
> Any chance I can use a MySQL database table to do a lookup to return the url to rewrite to? Let me explain the situation first... Magento utilizes rewrite rules heavily but does the majority of the rewrite within the Front controller. The problem is...

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 2:26am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/2 "2010-08-24T02:26:50Z")

</div>

you could cron a dump of the database into nginx rewrite rules, and  
then reload nginx…

otherwise, mysql lookups will block, and most nginx modules are  
designed with non-blocking in mind

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 2:44am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/3 "2010-08-24T02:44:24Z")

</div>

Hi,

> Any chance I can use a MySQL database table to do a lookup to return the  
> url to rewrite to?

Yes, you can do this with ngx\_drizzle combined with ngx\_lua.

agentzh will probably reply with working example soon enough 😉

Best regards,  
Piotr S. \< [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) \>

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 4:14am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/4 "2010-08-24T04:14:33Z")

</div>

On Tue, Aug 24, 2010 at 8:25 AM, Michael S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
wrote:

> otherwise, mysql lookups will block, and most nginx modules are  
> designed with non-blocking in mind

mysql lookups do not necessarily block if you do not use libmysql.

Our ngx\_drizzle module can talk with mysql 5.1+ without blocking the  
nginx process and it’s been heavily used in production. See  
[GitHub - openresty/drizzle-nginx-module: an nginx upstream module that talks to mysql and drizzle by libdrizzle · GitHub](http://github.com/chaoslawful/drizzle-nginx-module) for more details.

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 5:41am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/5 "2010-08-24T05:41:46Z")

</div>

You are a crazy, crazy guy agentzh! 🙂

On Mon, Aug 23, 2010 at 8:32 PM, agentzh [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> > Magento utilizes rewrite rules heavily but does the majority of the  
> > rewrite within the Front controller. The problem is that it is VERY  
> > in-efficient and a store with more than 10,000 products is severely  
> > delay. But, all information is referenced in a table ‘core\_url\_rewrite’  
> > so my though it so do the lookup first in Nginx and hand off to Magento  
> > with the non-seo url.

You could also whip up a quick little PHP script to do the  
pre-processing for you, not bootstrapping Magento, but simply db  
querying, and add memcache or apc or another caching mechanism,  
without having to go this deep into development modules and having to  
compile a bunch of external modules to get this to work 🙂

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 5:33am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/6 "2010-08-24T05:33:34Z")

</div>

On Tue, Aug 24, 2010 at 8:22 AM, btafoya [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> with the non-seo url.

Here’s a sample configuration that has been tested on my laptop.

Consider you seo uri is /baz, the true uri is /foo/bar. And I have the  
following table in my local mysql “test” database:

create table my\_url\_map(id serial, url text, seo\_url);  
insert into my\_url\_map(url, seo\_url)values(‘/foo/bar’, ‘/baz’);

And I build my nginx 0.8.41 this way:

```
./configure \
            --add-module=/path/to/ngx_devel_kit \
            --add-module=/path/to/set-misc-nginx-module \
            --add-module=/path/to/ngx_http_auth_request_module-0.2 \
            --add-module=/path/to/echo-nginx-module \
            --add-module=/path/to/lua-nginx-module \
            --add-module=/path/to/drizzle-nginx-module \
            --add-module=/path/to/rds-json-nginx-module

```

Also, I have lua 5.1.4 and the lua-yajl library installed to my system.

And here’s the central part in my nginx.conf:

```
upstream backend {
    drizzle_server 127.0.0.1:3306 dbname=test
         password=some_pass user=monty protocol=mysql;
    drizzle_keepalive max=300 mode=single overflow=ignore;
}

lua_package_cpath '/path/to/your/lua/yajl/library/?.so';

server {
      ...

    location /conv-mysql {
        internal;
        set_quote_sql_str $seo_uri $query_string; # to prevent sql 

```

injection  
drizzle\_query “select url from my\_url\_map where  
seo\_url=$seo\_uri”;  
drizzle\_pass backend;  
rds\_json on;  
}

```
    location /conv-uid {
        internal;
        content_by_lua_file 'html/foo.lua';
    }

    location /jump {
        internal;
        rewrite ^ $query_string? redirect;
    }

    # your SEO uri
    location /baz {
        set $my_uri $uri;
        auth_request /conv-uid;

        echo_exec /jump $my_uri;
    }

```

Then let’s access /baz from the client side:

```
$ curl -i localhost:1984/baz
HTTP/1.1 302 Moved Temporarily
Server: nginx/0.8.41 (without pool)
Date: Tue, 24 Aug 2010 03:28:42 GMT
Content-Type: text/html
Content-Length: 176
Location: http://localhost:1984/foo/bar
Connection: keep-alive

<html>
<head><title>302 Found</title></head>
<body bgcolor="white">
<center><h1>302 Found</h1></center>
<hr><center>nginx/0.8.41 (without pool)</center>
</body>
</html>

```

We can optimize this further by introducing ngx\_srcache, ngx\_memc and  
memcached clusters with connection pooling to cache our mysql result  
sets. But I’d keep this sample config minimal for now 🙂 If anybody is  
interested, just follow up and ask here 🙂

Note that 0.8.42+ won’t work for this config, and the request will  
hang for reasons that I don’t know yet. We do not have the resource to  
track all the aggressive changes in recent nginx core and that’s no  
fun at all.

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 5:56am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/7 "2010-08-24T05:56:59Z")

</div>

On Tue, Aug 24, 2010 at 11:40 AM, Michael S. [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
wrote:

> You are a crazy, crazy guy agentzh! 🙂

🙂

> You could also whip up a quick little PHP script to do the  
> pre-processing for you, not bootstrapping Magento, but simply db  
> querying, and add memcache or apc or another caching mechanism,  
> without having to go this deep into development modules and having to  
> compile a bunch of external modules to get this to work 🙂

Yup, sure we can do that. Just depend on how fast you want to achieve  
and how much resource you want to spend on those services 😉

On my laptop, nginx + php-fpm can only get about 4000 r/s for a hello  
world php script and once you start to connect to a database and do  
some simple calculations, oh well…

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 5:42am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/8 "2010-08-24T05:42:56Z")

</div>

On Tue, Aug 24, 2010 at 11:32 AM, agentzh [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> Here’s a sample configuration that has been tested on my laptop.

[snip]

> Connection: keep-alive

Just a quick note regarding performance: on my ThinkPad T400 (Core2Duo  
T9600) laptop, with only one nginx worker, and both ab and mysqld  
running on the same machine, and no memcached and no result set  
caching we get:

```
Server Software: nginx/0.8.41
Server Hostname: localhost
Server Port: 1984

Document Path: /baz
Document Length: 176 bytes

Concurrency Level: 10
Time taken for tests: 1.812 seconds
Complete requests: 10000
Failed requests: 0
Write errors: 0
Non-2xx responses: 10000
Total transferred: 3900000 bytes
HTML transferred: 1760000 bytes
Requests per second: 5518.47 [#/sec] (mean)
Time per request: 1.812 [ms] (mean)
Time per request: 0.181 [ms] (mean, across all concurrent 

```

requests)  
Transfer rate: 2101.76 [Kbytes/sec] received

And with ngx\_memc + ngx\_srcache enabled, we can achieve 6200+ r/s with  
the same settings.

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 6:04am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/9 "2010-08-24T06:04:00Z")

</div>

On Tue, Aug 24, 2010 at 11:32 AM, agentzh [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> ```
> location /conv-uid {
> internal;
> content_by_lua_file 'html/foo.lua';
> }
> 
> ```

Sorry, forgot to paste the contents of my html/foo.lua here, and it’s  
the essential glue:

```
-- foo.lua
local yajl = require('yajl')
local seo_uri = ngx.var.my_uri
local res = ngx.location.capture('/conv-mysql?' .. seo_uri)
if (res.status ~= ngx.HTTP_OK) then
    ngx.throw_error(res.status)
end
res = yajl.to_value(res.body)
if (not res or not res[1] or not res[1].url) then
    ngx.throw_error(ngx.HTTP_INTERNAL_SERVER_ERROR)
end
ngx.var.my_uri = res[1].url;

```

This piece of Lua code is mostly self-explanatory if you do know a  
little Lua 😉

Tests have shown that enabling LuaJIT 2.0 here won’t make things  
noticeably faster because we do not have enough Lua for JIT’ing 😉

Have fun!  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 6:04am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/10 "2010-08-24T06:04:00Z")

</div>

That is more along the lines of what I was thinking Mike, but have no  
experience doing so with Magento. There is always a first for everything  
though!

Thanks!

Posted at Nginx Forum:

> **[Re: Rewrite lookup via MySQL table](https://forum.nginx.org/read.php?2,122977,123018#msg-123018)**
>
> Any chance I can use a MySQL database table to do a lookup to return the url to rewrite to? Let me explain the situation first... Magento utilizes rewrite rules heavily but does the majority of the rewrite within the Front controller. The problem is...

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 6:12am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/11 "2010-08-24T06:12:47Z")

</div>

sounds like it’s just some db queries. so really you’ll intercept it  
in nginx, pass it off to php, and then redirect those requests (is it  
a 301 redirect?) to the proper URL.

it’s a very basic PHP/mysql setup. always start simple, so add in  
caching once things work…

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 6:06am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/12 "2010-08-24T06:06:56Z")

</div>

Very cool! I am going to try some pre-processing in PHP first and see  
how that goes. If that does not work, I will look into your suggestion.

Thanks VERY much!

Posted at Nginx Forum:

> **[Re: Rewrite lookup via MySQL table](https://forum.nginx.org/read.php?2,122977,123019#msg-123019)**
>
> Any chance I can use a MySQL database table to do a lookup to return the url to rewrite to? Let me explain the situation first... Magento utilizes rewrite rules heavily but does the majority of the rewrite within the Front controller. The problem is...

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 8:42pm UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/13 "2010-08-24T20:42:50Z")

</div>

Forgot to show what the target path is now:

[http://fcdev.premasolutions.com/catalog/product/view/id/9746/category/743](http://fcdev.premasolutions.com/catalog/product/view/id/9746/category/743)

## btafoya Wrote:

> loop. The handler I wrote is index\_loader.php.
> 
> ```
> location ~ \.php/ { ## Forward paths like
> 
> ```
> 
> /js/index.php/x.js to relevant handler  
> rewrite ^(.\*.php)/ $1 last;  
> }
> 
> How do I go about redirecting all .html traffic to  
> the new handler without interfering with the  
> Magento front controller?
> 
> Thanks again for everyone’s help!

Posted at Nginx Forum:  
[http://forum.nginx.org/read.php?2,122977,123234#msg-123234](http://forum.nginx.org/read.php?2,122977,123234#msg-123234)

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 24, 2010, 8:29pm UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/14 "2010-08-24T20:29:22Z")

</div>

I have the php code working to redirect correctly, but have run into a  
obstacle I don’t know how to resolve…

The is the incoming URL, or one of them…

[http://fcdev.premasolutions.com/alice-in-wonderland-movie-deluxe-alice-adult-costume.html](http://fcdev.premasolutions.com/alice-in-wonderland-movie-deluxe-alice-adult-costume.html)

I have tried to intercept it but just created a loop. The handler I  
wrote is index\_loader.php.

```
location ~ \.html/ { ## Forward
    rewrite ^(.*)/ index_loader.php?source_url=$1 last;
}

location @handler { ## Magento uses a common front handler
    rewrite / /index.php;
}

location ~ \.php/ { ## Forward paths like /js/index.php/x.js to

```

relevant handler  
rewrite ^(.\*.php)/ $1 last;  
}

How do I go about redirecting all .html traffic to the new handler  
without interfering with the Magento front controller?

Thanks again for everyone’s help!

Posted at Nginx Forum:

> **[Re: Rewrite lookup via MySQL table](https://forum.nginx.org/read.php?2,122977,123227#msg-123227)**
>
> Any chance I can use a MySQL database table to do a lookup to return the url to rewrite to? Let me explain the situation first... Magento utilizes rewrite rules heavily but does the majority of the rewrite within the Front controller. The problem is...

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 25, 2010, 7:09am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/15 "2010-08-25T07:09:52Z")

</div>

email me off list i can try to help - mike503@gmail

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 25, 2010, 7:50am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/16 "2010-08-25T07:50:50Z")

</div>

On Tue, Aug 24, 2010 at 11:32 AM, agentzh [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> Note that 0.8.42+ won’t work for this config, and the request will  
> hang for reasons that I don’t know yet. We do not have the resource to  
> track all the aggressive changes in recent nginx core and that’s no  
> fun at all.

This is a false alarm. My config pasted above should work for nginx  
0.8.42+ (as far as nginx 0.8.49 in fact) 🙂

I was fooled by my local use of ngx\_srcache (in subrequests) which was  
known to hang due to a “regression” in nginx 0.8.42+ explained below:

```
http://forum.nginx.org/read.php?29,103078,103078

```

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 28, 2010, 2:55pm UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/17 "2010-08-28T14:55:31Z")

</div>

First of all, this thread contains a lot of information! I’m in the  
process of implementing something similar and it has helped me a ton.

I was just wondering, as long as you’re using Lua anyway, why not do  
away with Drizzle completely, and just make Lua connect to the database  
to retrieve the correct URL? Is it because Drizzle will keep one  
connection to the database per nginx-instance, so there is less  
overhead, compared to Lua which would have to connect and disconnect to  
the database for every request?

Posted at Nginx Forum:

> **[Re: Rewrite lookup via MySQL table](https://forum.nginx.org/read.php?2,122977,124586#msg-124586)**
>
> Any chance I can use a MySQL database table to do a lookup to return the url to rewrite to? Let me explain the situation first... Magento utilizes rewrite rules heavily but does the majority of the rewrite within the Front controller. The problem is...

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 30, 2010, 5:21am UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/18 "2010-08-30T05:21:41Z")

</div>

On Sat, Aug 28, 2010 at 8:54 PM, relix [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> First of all, this thread contains a lot of information! I’m in the  
> process of implementing something similar and it has helped me a ton.
> 
> I was just wondering, as long as you’re using Lua anyway, why not do  
> away with Drizzle completely, and just make Lua connect to the database  
> to retrieve the correct URL?

Lua’s database library will block the nginx process which is  
unacceptable for real world applications. ngx\_drizzle is served as a  
non-blocking mysql driver component for Lua here (of cause, it can  
also work with ngx\_http\_js\_module and others).

> Is it because Drizzle will keep one  
> connection to the database per nginx-instance, so there is less  
> overhead,

Actually ngx\_drizzle can keep connection pools to mysql backends per  
nginx instance. One db connection per process is the suboptimal  
PHP/Perl way of doing things and it can waste a lot of database  
resources.

> compared to Lua which would have to connect and disconnect to  
> the database for every request?

MUCH more than just that 😉

Cheers,  
-agentzh

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 30, 2010, 12:33pm UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/19 "2010-08-30T12:33:39Z")

</div>

I am new to nginx,  
but, as far as I know,  
nginx is known to be a fast light webserver.

So, having each rewrite to be looked up in a mysql table seems  
something very slow idea.  
rewrite is something to be done on the fly, as fast as possible. and  
putting a mysql layer in this process seems not a good idea.

---

<div class="post-metadata">

**Author:** ![btafoya](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/b/258eb7/32.png) [@btafoya](https://www.ruby-forum.com/u/btafoya)\
**Post date:** [August 30, 2010, 3:05pm UTC](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766/20 "2010-08-30T15:05:17Z")

</div>

Ah, thanks, that does make sense!

## agentzh Wrote:

> > to retrieve the correct URL?  
> > connection to the database per nginx-instance,
> 
> * * *
> 
> nginx mailing list  
> [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid)  
> [mailman.nginx.org archive](http://nginx.org/mailman/listinfo/nginx)

Posted at Nginx Forum:  
[http://forum.nginx.org/read.php?2,122977,125201#msg-125201](http://forum.nginx.org/read.php?2,122977,125201#msg-125201)

[Next page](https://www.ruby-forum.com/t/rewrite-lookup-via-mysql-table/193766.md?page=2)