How-to Lighttpd, PHP, MySQL and Eaccelerator

Quote:

---

Originally Posted by BrandonSk

Now, server starts and works. But was this the right thing to do? Am I missing something because of that? I suppose https :)

Cheers,
Brano.

---

SSL isn't configured in the config file, for that you need to do some things yourself:)

You can make a self signed pem file yourself like this:

Quote:

---

openssl req -new -x509 -keyout server.pem -out server.pem -days 365 -nodes

---

you need openssl installed for that

now for the lighttpd config file (just add it somewhere):

Quote:

---

$SERVER["socket"] == ":443" {
#### SSL engine
ssl.engine = "enable"
ssl.pemfile = "/opt/etc/lighttpd/server.pem"
server.document-root = "/harddisk/mnt/www"
}

---

This creates a separate socket with ssl
this way your site works via http and https

try it out on my site, I've been running that for a month or so: https://wpte.kicks-ass.net/

Now obviously browsers say its unsafe since it's not signed by a known company.
You can still import it for general accepted certificates tho, but for windows you need something like a .der file
this is how you convert the pem file to a der file:

Quote:

---

openssl x509 -in server.pem -inform PEM -out server.der -outform DER

---

I'm going to update this how-to soon btw, since I found out I have 3 major errors in it :)

Hi!
I just installed everything according to the 1.st post.
I got 3 problems, 2 of them are solved in this forum, they were:
1.) When using the ipkg eaccelerator, server gives internal error 500 when the extension of a file is *.php, but it works well with *.html.
solution: do the manual install.
2.) couldn't connect to the mysql, something about authorization error
solution: delete or comment the 20.th line of my.cnf that says password

And finaly the one i cannot find anywhere.
3.) I dumped a database from my wamp installation.
mysqldump kruh > kruh.sql;
i moved this file to my router, executed it with
source kruh.sql;
everything works fine, i am able to run selects, for instance select(*) from users; gives me 4.
But i cannot get this to work in php.
My code:

Code:

---

$db_host = 'localhost';
$db_user = 'bady';
$db_pass = '';
$db_data = 'kruh';
// ---------------------------------------------------
$link=mysql_connect($db_host,$db_user,$db_pass);
if(!$link) die("unable to connect: ".mysql_error());
mysql_select_db($db_data,$link);
$x=mysql_result(mysql_query("select count(*) from users",$link),0);
echo $x;

---

This returns nothing :(
If i change the select to "select NOW() from dual" it works fine, so im connected to DB. What am i doing wrong?

EDIT: this solved the problem:
typed this in mysql console: grant all privileges on *.* to ''@'localhost';

Hi,

I've created .pem and .der files with openssl after which I've added lines proposed by Wpte to lighttpd.conf file. Restart of lighttpd and no monit about certificate under https:// address. Should I have 443 port forwarded or what?

Quote:

---

Originally Posted by cichy

Hi,

I've created .pem and .der files with openssl after which I've added lines proposed by Wpte to lighttpd.conf file. Restart of lighttpd and no monit about certificate under https:// address. Should I have 443 port forwarded or what?

---

What exactly happens?
you can see your webpages in https?
or isn't it encrypted at all?


@badyto
I know there are serious problems with the how-to at the moment due to updated ipkg packages.
Unfortunately I haven't found the time to fully test my findings.
The updated how-to will be shorter and easier to understand

Basically when i try to connect thru https:// i don't get a reply, no 501 or 404 error just time out in FireFox and IE 8. I've tried to add this

Code:

---

$HTTP["scheme"] == "https" {
    server.document-root        = "/www/servers/www.example.org/secure/"
 }

---

but it didn't helped at all. Thru http:// I can use my site but not thru https://

Quote:

---

Originally Posted by cichy

Basically when i try to connect thru https:// i don't get a reply, no 501 or 404 error just time out in FireFox and IE 8. I've tried to add this

Code:

---

$HTTP["scheme"] == "https" {
    server.document-root        = "/www/servers/www.example.org/secure/"
 }

---

but it didn't helped at all. Thru http:// I can use my site but not thru https://

---

You mentioned port forwarding as well?
if you're in the local network of your router no port-forwarding is needed

but for external access you can add a basic iptables rule like this to your firewall:

Quote:

---

iptables -A INPUT -p tcp --dport 443 -j ACCEPT

---

in the end, the config settings I wrote in my previous post should be all to make https work.
I've double checked it in my config

Added iptables rule and didn't helped still can't use SSL connections. Maybe something is wrong with my .conf file, could you look into it?

Edit: This is new when I've entered local IP - https://192.168.1.1 a certificate notification appeared so it works, but why not on my external IP?

Quote:

---

Originally Posted by cichy

Added iptables rule and didn't helped still can't use SSL connections. Maybe something is wrong with my .conf file, could you look into it?

Edit: This is new when I've entered local IP - https://192.168.1.1 a certificate notification appeared so it works, but why not on my external IP?

---

Good to know it works locally:)

Just as the "server.port = 8081" the ssl socket should bind to every network interface.
if it's not the modem that requires an extra firewall setting you might try these config changes:

Quote:

---

$SERVER["socket"] == "0.0.0.0:443"{
...


$SERVER["socket"] == "127.0.0.1:443"{
...

---

not sure if they work, but it's worth trying


I do think it's something with the firewall tho, since:

Quote:

---

x.x.x.x isn't responding on port 443 (https).

---

(as moderator I can see your ip, no worries;))

This didn't helped either, still no ssl connections but on local IP it works (sic!). Also I've tried some of this tutorial on port changing.

Code:

---

nvram set http_lanport=8080
nvram commit

---

did that, restarted router and sure web-gui works fine but no website on my router (port changed from 8081 to 80 like in tut). Also I removed 1 rule from iptables which I thought is not usefull any more, this was:

Code:

---

iptables -t nat -A VSERVER -p tcp -m tcp --dport 80 -j DNAT --to-destination 192.168.1.1:8081        # przekierowanie www

---

Any ideas what's wrong, lighty is working according to htop.

Edit: Https:// connection with address https://ex.ip:8081 works but why not with https://ex.ip ?

Quote:

---

Originally Posted by cichy

Edit: Https:// connection with address https://ex.ip:8081 works but why not with https://ex.ip ?

---

I know that lighttpd is a bit strict when it comes to ports...
have you tried to set the standard port to 80 + my config code?
if that works, it might be a lighttpd bug, or setting I'm not aware of:confused:


https://ex.ip:8081 doesn't work for me here

Ok so I've reinstalled lighttpd (was not worki with/without ssl) now http connections work like they should. Setting lan_httport to 8080 did the trick but i had to manually forward all incoming http request to 192.168.1.1:80 to make it work after changing server.port to 80. Still no connection thru ssl.

On internal ip (192.168.1.1) https works like a charm.

Quote:

---

Originally Posted by cichy

Ok so I've reinstalled lighttpd (was not worki with/without ssl) now http connections work like they should. Setting lan_httport to 8080 did the trick but i had to manually forward all incoming http request to 192.168.1.1:80 to make it work after changing server.port to 80. Still no connection thru ssl.

On internal ip (192.168.1.1) https works like a charm.

---

Right... so how do you have the regular iptables for forwarding the webserver?

Quote:

---

iptables -t nat -A VSERVER -p tcp -m tcp --dport 8081 -j DNAT --to-destination 192.168.1.1:8081

---

?

Are you sure your modem is set up for forwarding https?

I've attached my post-firewall file, also in error log i found

Code:

---

2010-09-09 21:39:26: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:40:12: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:42:24: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:42:27: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request

---

i think this is the problem.

Quote:

---

Originally Posted by cichy

I've attached my post-firewall file, also in error log i found

Code:

---

2010-09-09 21:39:26: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:40:12: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:42:24: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request
2010-09-09 21:42:27: (connections.c.299) SSL: 1 error:1407609C:SSL routines:SSL23_GET_CLIENT_HELLO:http request

---

i think this is the problem.

---

that indeed is a problem...
I'm not sure if I'm getting this, but lighttpd says it fails to do https because the browser makes a http request?:confused:

The weird thing is, everything is set up according to the lighty wiki: http://redmine.lighttpd.net/projects.../wiki/Docs:SSL

is your lighttpd package up to date?
my lighttpd -V output:

Code:

---

lighttpd -V
lighttpd/1.4.28 (ssl) - a light and fast webserver
Build-Date: Aug 22 2010 18:32:55

Event Handlers:

        + select (generic)
        + poll (Unix)
        - rt-signals (Linux 2.4+)
        + epoll (Linux 2.6)
        - /dev/poll (Solaris)
        - kqueue (FreeBSD)
        - libev (generic)

Network handler:

        + writev
        + mmap support

Features:

        + IPv6 support
        + zlib support
        + bzip2 support
        + crypt support
        + SSL Support
        + PCRE support
        + mySQL support
        + LDAP support
        + memcached support
        - FAM support
        + LUA support
        + xml support
        + SQLite support
        - GDBM support

---

default ipkg updating and upgrading if you're not aware of it:

Quote:

---

ipkg update
ipkg upgrade

---

Quote:

---

Originally Posted by badyto

Hi!
I just installed everything according to the 1.st post.

---

Hello Badyto, I updated the how-to
I tested it on my own spare router, and it should work without any problems:)

And this is mine lighttpd -V result:

Code:

---

lighttpd/1.4.28 (ssl) - a light and fast webserver
Build-Date: Aug 22 2010 18:32:55

Event Handlers:

        + select (generic)
        + poll (Unix)
        - rt-signals (Linux 2.4+)
        + epoll (Linux 2.6)
        - /dev/poll (Solaris)
        - kqueue (FreeBSD)
        - libev (generic)

Network handler:

        + writev
        + mmap support

Features:

        + IPv6 support
        + zlib support
        + bzip2 support
        + crypt support
        + SSL Support
        + PCRE support
        + mySQL support
        + LDAP support
        + memcached support
        - FAM support
        + LUA support
        + xml support
        + SQLite support
        - GDBM support

---

As You can see they are identical. Everything is up2date so no mistakes because of wrong software. I've also studied the site You gave here and without a doubt it should work but it ain't.

Need some help.

I've updated PHP to latest version and recompiled eAccelerator from scratch but this didn't helped at all, fcgi process dies in a second after lighttpd restart.

Code:

---

2010-12-29 21:33:01: (log.c.166) server started
2010-12-29 21:33:09: (mod_fastcgi.c.2568) unexpected end-of-file (perhaps the fastcgi process died): pid: 3751 socket: unix:/tmp/php-fcgi.sock-3
2010-12-29 21:33:09: (mod_fastcgi.c.3309) child exited, pid: 3751 status: 254
2010-12-29 21:33:09: (mod_fastcgi.c.3356) response not received, request sent: 818 on socket: unix:/tmp/php-fcgi.sock-3 for /index.php?, closing connection
2010-12-29 21:33:11: (mod_fastcgi.c.2568) unexpected end-of-file (perhaps the fastcgi process died): pid: 3752 socket: unix:/tmp/php-fcgi.sock-3
2010-12-29 21:33:11: (mod_fastcgi.c.3356) response not received, request sent: 818 on socket: unix:/tmp/php-fcgi.sock-3 for /index.php?, closing connection
2010-12-29 22:45:25: (mod_fastcgi.c.1734) connect failed: Connection refused on unix:/tmp/php-fcgi.sock-3
2010-12-29 22:45:25: (mod_fastcgi.c.3027) backend died; we'll disable it for 1 seconds and send the request to another backend instead: reconnects: 0 load: 1
2010-12-29 22:45:28: (mod_fastcgi.c.2568) unexpected end-of-file (perhaps the fastcgi process died): pid: 3756 socket: unix:/tmp/php-fcgi.sock-3
2010-12-29 22:45:28: (mod_fastcgi.c.3356) response not received, request sent: 818 on socket: unix:/tmp/php-fcgi.sock-3 for /index.php?, closing connection
2010-12-29 23:02:51: (server.c.1512) server stopped by UID = 0 PID = 3767
2010-12-29 23:02:53: (log.c.166) server started
2010-12-29 23:21:41: (server.c.1512) server stopped by UID = 0 PID = 3798
2010-12-29 23:22:44: (log.c.166) server started
2010-12-29 23:23:08: (mod_fastcgi.c.1734) connect failed: Connection refused on unix:/tmp/php-fcgi.sock3808-7
2010-12-29 23:23:08: (mod_fastcgi.c.3027) backend died; we'll disable it for 1 seconds and send the request to another backend instead: reconnects: 0 load: 1
2010-12-29 23:23:13: (mod_fastcgi.c.2568) unexpected end-of-file (perhaps the fastcgi process died): pid: 3820 socket: unix:/tmp/php-fcgi.sock3808-7
2010-12-29 23:23:13: (mod_fastcgi.c.3309) child exited, pid: 3820 status: 254
2010-12-29 23:23:13: (mod_fastcgi.c.3356) response not received, request sent: 818 on socket: unix:/tmp/php-fcgi.sock3808-7 for /index.php?, closing connection

---

Maybe some clues what to do?

Quote:

---

Originally Posted by cichy

Need some help.

I've updated PHP to latest version and recompiled eAccelerator from scratch but this didn't helped at all, fcgi process dies in a second after lighttpd restart.

Maybe some clues what to do?

---

download some fresh sources if you haven't already.
Also, try to delete the old eaccelerator first before you make install:

Quote:

---

rm /opt/lib/php/extensions/eaccelerator.so

---

not sure, but over time ipkg packages changed quite a bit
In fact, they solved a lot of workarounds I wrote here in the how-to
so, I had to remove the workarounds again:p
if all else fails, you might want to try the ipkg version of eaccelerator

I've tried the ipkg source version first but no effect still errors and php dies after a sec. Now I'm recompiling eaccelerator from scratch from latest svn build, maybe this will help. Also, strange thing, when i've disabled eaccelerator in eaccelerator.ini and disabled mod_fcgi in lighttpd.conf I can't launch any file from server i.e *.html file, I'm getting 403 - access denied.

Quote:

---

Originally Posted by cichy

I've tried the ipkg source version first but no effect still errors and php dies after a sec. Now I'm recompiling eaccelerator from scratch from latest svn build, maybe this will help. Also, strange thing, when i've disabled eaccelerator in eaccelerator.ini and disabled mod_fcgi in lighttpd.conf I can't launch any file from server i.e *.html file, I'm getting 403 - access denied.

---

huh thats weird:confused:
have you checked the extra config files in /opt/etc/lighttpd/conf.d ?

Yep there're 2 files 1st default (empty) and 2nd fcgi with part from lighttpd.conf of mod_fcgi. Something like that:

Code:

---

fastcgi.server            = ( ".php" =>
                              ( "localhost" =>
                                (
                                  "socket" => "/tmp/php-fcgi.sock",
                                  "bin-path" => "/opt/bin/php-fcgi",
                                "max-procs" => 1,
                                  "bin-environment" => (
                                        "PHP_FCGI_CHILDREN" => "3",
                                          "PHP_FCGI_MAX_REQUESTS" => "4000"
                                  )
                                )
                              )
                            )

---

After deleting eaccelerator.so i could query php -i, and now it worked bot warned me about missing module of eaccelerator, maybe this recompiling helps :)

Quote:

---

Originally Posted by cichy

Yep there're 2 files 1st default (empty) and 2nd fcgi with part from lighttpd.conf of mod_fcgi. Something like that:

After deleting eaccelerator.so i could query php -i, and now it worked bot warned me about missing module of eaccelerator, maybe this recompiling helps :)

---

sounds promising:)

that fcgi file already configures your php, so you shouldn't reactivate it in the main config file (like in the old how to)

Did something like this

Code:

---

# Configuration from other optware packages
#include_shell "cat /opt/etc/lighttpd/conf.d/*.conf"

---

So it uses part from .conf file without external modules/files. Now waiting to configure it once again because of missing module :( Hope this will help :)

And recompiling didn't helped at all, still no fcgi engine running and got this error:

Code:

---

(plugin.c.131) Cannot load plugin mod_fastcgi more than once, please fix your config (we may not accept such configs in future releases)

---

Maybe some clues how to solve this issue.

Quote:

---

Originally Posted by cichy

And recompiling didn't helped at all, still no fcgi engine running and got this error:

Code:

---

(plugin.c.131) Cannot load plugin mod_fastcgi more than once, please fix your config (we may not accept such configs in future releases)

---

Maybe some clues how to solve this issue.

---

sounds like you have the conf.d fcgi script included, and uncommented it in the main config file.

so you need to uncomment the "mod_fastcgi" in the server.modules list.
because in the php-fcgi.conf file it says:

Quote:

---

server.modules += ( "mod_fastcgi" )

---

:)

So where should i uncomment it coz frankly i dont understand what you mean. Where (which file) and what shoul i uncomment? :)

Quote:

---

Originally Posted by cichy

So where should i uncomment it coz frankly i dont understand what you mean. Where (which file) and what shoul i uncomment? :)

---

Sorry:p
In the beginning of the main (lighttpd.conf) config file.
Just make sure you don't include the php-fcgi module twice:)

Take a look at my lighttpd.conf file, maybe I'm doing something wrong. Also when i type php -i in console it gives me eaccelerator error.

Code:

---

PHP Fatal error:  [eAccelerator] eAccelerator 0.9.6.1 can not be loaded twice in Unknown on line 0

---

When I disable fcgi support in conf file i can view html files but that's all.

Quote:

---

Originally Posted by cichy

Take a look at my lighttpd.conf file, maybe I'm doing something wrong. Also when i type php -i in console it gives me eaccelerator error.

Code:

---

PHP Fatal error:  [eAccelerator] eAccelerator 0.9.6.1 can not be loaded twice in Unknown on line 0

---

When I disable fcgi support in conf file i can view html files but that's all.

---

these are the fixed workarounds I was talking about:)
the current eAccelerator ipkg package (and my how-to as well) adds a seperate config file: /opt/etc/php.d/eaccelerator.ini

so if you added that, either way you should remove the config file, or delete the part about eAccelerator in /opt/etc/php.ini :p

This helped, deleting from php.ini I mean. Maybe tutorial should be updated with this info :)

Quote:

---

Originally Posted by cichy

This helped, deleting from php.ini I mean. Maybe tutorial should be updated with this info :)

---

Well I can't keep all the changed things in the how-to... I mean, it would make things more complicated to follow I think:)

Probably Yes, either way thanks for your help :)

I installed lighttpd and whenever I start it, immediately closes without any warnings. If I type lighttpd without any option :

Code:

---

[admin@WL-00221561958C root]$ lighttpd
2011-06-11 22:55:25: (server.c.588) No configuration available. Try using -f option.

---

so its running. I've tried the original config file and the same thing. I've even deliberately inserted a mistake in the config file and it reported the problem correctly.

So why isn't the lighttpd process keep running ?

edit: Oh .. and I did use lighttpd successfully in the past on the same setup.

Quote:

---

Originally Posted by expert_vision

I installed lighttpd and whenever I start it, immediately closes without any warnings. If I type lighttpd without any option :

Code:

---

[admin@WL-00221561958C root]$ lighttpd
2011-06-11 22:55:25: (server.c.588) No configuration available. Try using -f option.

---

so its running. I've tried the original config file and the same thing. I've even deliberately inserted a mistake in the config file and it reported the problem correctly.

So why isn't the lighttpd process keep running ?

edit: Oh .. and I did use lighttpd successfully in the past on the same setup.

---

uhm... according to that line it's not running, in fact it stops because you didn't supply a configuration file which is done by using the -f parameter:)

Quote:

---

/opt/sbin/lighttpd -f /opt/etc/lighttpd/lighttpd.conf

---

should do the trick

No .. it didn't start even with the config file.
Anyway I looked in the error.log and found this

Code:

---

2011-06-11 23:40:14: (log.c.166) server started
2011-06-11 23:40:14: (fdevent_linux_sysepoll.c.134) epoll_create failed (Function not implemented), try to set server.event-handler = "poll" or "select"
2011-06-11 23:40:14: (fdevent.c.41) event-handler linux-sysepoll failed, try to set server.event-handler = "poll" or "select"
2011-06-11 23:40:14: (server.c.1112) fdevent_init failed

---

I did just that in the config file and
lighttpd server is running. :D

I have no idea what server.event-handler = "poll" is .. and I don't think I want to know.

Quote:

---

Originally Posted by expert_vision

I did just that in the config file and
lighttpd server is running. :D

I have no idea what server.event-handler = "poll" is .. and I don't think I want to know.

---

uhm... the how-to tells you to do so:

Quote:

---

This is for fixing the excluded e-poll:

Quote:

---

line 51: change

# server.event-handler = "freebsd-kqueue" # needed on OS X

into

server.event-handler = "poll" # needed on OS X

---

---

back in the days epoll was included in the ipkg package, not anymore tho... and since our firmware doesn't have it we need poll:)
You don't want to know what it's for? like it would be some kind of dirty thing:o
To explain it in short: it changes the way how an incoming web-request is trown at the event-handler. epoll is supposed to be more efficient, but it only shows a real difference when your webserver is accessed a lot:)

I know it is in the how-to, but since I used lighttpd before without epoll, I thought it wasn't mandatory, and apparently that changed in the new version.
And hey .. thanks for still supporting this how-to!

Quote:

---

Originally Posted by expert_vision

I know it is in the how-to, but since I used lighttpd before without epoll, I thought it wasn't mandatory, and apparently that changed in the new version.
And hey .. thanks for still supporting this how-to!

---

ah cool:)
Yeah well, need to keep the fans happy right:p;)

I tried to install mysql5 from the ipkg feed today...
after some user rights stuff I managed to get it running perfectly fine, the downside however is that the php-mysql plugin doesn't seem to be compatible with the newer mysql5:o

so no mysql5 support in this how-to for now:)

Hi,

first of all, great tutorial, easy to follow and set-up, everything working fine.

I just want to ask if there is a way to run php 5.3 on wl500gp. Is there a guide to do this? I am not very skilled in unix systems :(
Additionaly, is it yet possible to run mysql 5 on this box?

I need this environment as the tool I am trying to install has these requirements.

Thanks