Having huge redirect list in .htaccess a Problem?_问答_开发者

I want to redirect every post 301 redirect, but I have over 3000 posts.

If I list

Redirect permanent /blog/2010/07/post.html http://new.blog.com/2010/07/23/post/
Redirect permanent /blog/2010/07/post1.html http://new.blog.com/2010/07/24/post1/
Redirect permanent /blog/2010/07/post2.html http://new.blog.com/2010/07/25/post2/
Redirect permanent /blog/2010/07/post3.html http://new.blog.com/2010/07/26/post3/
Redirect per......

for over 3000 url redirect command in .htaccess would this eat my server resource or cause some problem? Im not sure how .htaccess work but if the server is looking at these lists each time user requests for page, I would guess it will be a resource hog.

I can't use RedirectMatch because I added date variable in my new url. Do you have any other suggestions red开发者_StackOverflowirecting these posts? Or am I just fine?

Thanks!

I am not an Apache expert, so I cannot speak to whether or not having 3,000 redirects in .htaccess is a problem (though my gut tells me it probably is a bad idea). However, as a simpler solution to your problem, why not use mod_rewrite to do your redirects?

RewriteRule ^/blog/(.+)/(.+)/(.+).html$ http://new.blog.com/$1/$2/$3/ [R=permanent]

This uses a regex to match old URLs and rewrite them to new ones. The [R=permanent] instructs mod_rewrite to issue a 301 with the new URL instead of silently rewriting the request internally.

In your example, it looks like you've added the day of the post to the URL, which does not exist in the old URL. Since you obviously cannot use a regexp to divine the day an arbitrary post was made, this method may not work for you. If you can drop the day from the URL, then you're good to go.

Edit: The first time I read your question, I missed the last paragraph. ("I can't use RedirectMatch because I added date variable in my new url.") In this case, you can use mod_rewrite's RewriteMap to lookup the day component of a post.

You have two options:

Use a hashmap to perform fast lookups in a static file. This means all your old URLs will work, but any new posts cannot be accessed using the old URL scheme.
Use a script to grab the day.

In option one, create a file called posts.txt and put:

/yyyy/mm/pppp dd

...for each post where yyyy is the year of the post, mm is the month, and pppp is the post name (without the .html).

When you're done, run:

$ httxt2dbm -i posts.txt -o posts.map

Then we add to to the server/virtual server config: (Note the path is a filesystem path, not a URL.)

RewriteMap postday dbm:/path/to/file/posts.map
RewriteRule ^/blog/(.+)/(.+)/(.+).html$ http://new.blog.com/$1/$2/${postday:$1/$2/$3}/$3/ [R=permanent]

In option two, use pgm:/path/to/script/lookup.whatever as your RewriteMap. See the mod_rewrite documentation for more info about using a script.

Doing the lookup in mod_rewrite is better than just redirecting to a script which looks up the date and then redirects to the final destination because you should never redirect more than once. Issuing a 301 or 302 incurs a round trip cost, which increases the latency of your page load time.

If you have some way in code to determine the day of a post, you can generate the rewrite on the fly. You can setup a mod_rewrite pattern, something like .html and set up a front controller pattern to calculate the new url from the old and issue the 301 header.

With php as an example:

$_SERVER['REQUEST_URI']

will contain the requested url and

header("Location: http://new.blog.com/$y/$m/$d/$title/",TRUE,301);

will send a redirect.

That's... a lot of redirects. But the first thing I would tell you, and probably the only thing I can tell you without qualification, is that you should run some tests and see what the access times for your blog are like, and also look at the server's CPU and memory usage while you're doing it. If they're fairly low even with that giant list of redirects, you're okay as long as your blog doesn't experience a sudden increase in traffic. (I strongly suspect the 3000 rewrites will be slowing Apache down a lot, though)

That being said, I would second josh's suggestion of replacing the redirects with something dynamic. Like animuson said, if you're willing to drop the day from the URL, it'll be easy to set up a RewriteRule directive to handle the redirection. Otherwise, you could do it with a PHP script, or generally some code in whatever scripting language you (can) use. If you're using one of the popular blog engines, it probably contains code to do this already. Basically you could do something like

RewriteRule .* /blog/index.php

and just let the PHP script sort out which post was requested. It has access to the database so it'll be able to do that, and then you can either display the post directly from the PHP script, or to recover your original redirection behavior, you can send a Location header with the correct URL.

An alternative would be to use RewriteMap, which lets you write a RewriteRule where the target is determined by a program or file of your choice instead of being directly specified in the configuration file. As one option, you can specify a text file that contains the old and new URLs, and Apache will handle searching the file for the appropriate line for any given request. Read the documentation (linked above) for the full details. I will mention that this isn't used very often, and I'm not sure how much faster it would be compared to just having 3000 redirects.

Last tip: Apache can be significantly faster if you're able to move the configuration directives (like Redirect) into the server or virtual host configuration file, and disable reading of .htaccess entirely. I would guess that moving 3000 directives from .htaccess into the virtual host configuration could make your server considerably faster. But even moving the directives into the vhost config file probably wouldn't produce as much of a speedup as using a single RewriteRule.

It's never a good idea to make a massive list of Redirects. A better programming technique is to simply redirect the pages without that date variable then have a small PHP snippet that detects if it's missing and redirects to the URL with it included. The long list looks tacky and slows down Apache because it's checking that URL (any every other URL that might not even be affected by this) against each line. If it were only 5 or so, I'd say fine, but 3,000 is a definite NO.

Although I'm not a big fan of this method, a better choice would be to redirect all those URLs normally using a single match statement, redirecting them to the page without the date part, or with a dash or something, then include a small PHP snippet to check if the date is valid and if not, rewrite the path again to the correctly formed URL.

Honestly, if you didn't have that part there before, you don't need it now, and it will probably just confuse the search engines changing the URL for 3,000 posts. You don't really need a date in the URL, a good title is much more meaningful not only to users, but also to search engines, than a bunch of numbers.