Production Web server: Lighttpd: Difference between revisions

From Alpine Linux
(→‎Lighttpd advanced: fix max-ds and network write method command with sed..)
(3 intermediate revisions by the same user not shown)
Line 25: Line 25:


sed -i -r 's#.*server.stat-cache-engine.*=.*# server.stat-cache-engine = "fam"#g' /etc/lighttpd/lighttpd.conf
sed -i -r 's#.*server.stat-cache-engine.*=.*# server.stat-cache-engine = "fam"#g' /etc/lighttpd/lighttpd.conf
sed -i -r 's#.*server.pid-file.*=.*#server.pid-file      = "/run/lighttpd/lighttpd.pid"#g' /etc/lighttpd/lighttpd.conf


sed -i -r 's#\#.*server.event-handler = "linux-sysepoll".*#server.event-handler = "linux-sysepoll"#g' /etc/lighttpd/lighttpd.conf
sed -i -r 's#\#.*server.event-handler = "linux-sysepoll".*#server.event-handler = "linux-sysepoll"#g' /etc/lighttpd/lighttpd.conf
Line 222: Line 220:


== Lighttpd advanced ==
== Lighttpd advanced ==
Lighttpd has pretty good default settings, but a few might be tweaked if we need to respond to higher server loads. The more important area of tuning is simply enabling the advanced features of the recents Alpine Linux kernel we are using: Enable sys_epoll, sendfile and disable atime updates.


=== Lighttpd tunning for agressive load ===
=== Lighttpd tunning for agressive load ===
Line 229: Line 229:
<pre>
<pre>
<nowiki>
<nowiki>
checkset="";checkset=$(grep 'noatime' /etc/lighttpd/lighttpd.conf);[[ "$checkset" != "" ]] && echo listo || sed -i -r 's#.*{{{ server settings.*#{{{ server settings"\nserver.use-noatime = "enable"\n#g' /etc/lighttpd/lighttpd.conf
checkset="";checkset=$(grep 'noatime' /etc/lighttpd/lighttpd.conf);[[ "$checkset" != "" ]] && echo listo || sed -i -r 's#server settings.*#server settings"\nserver.use-noatime = "enable"\n#g' /etc/lighttpd/lighttpd.conf


rc-service lighttpd restart
rc-service lighttpd restart
Line 261: Line 261:


One reason that file descriptors get used up so quickly is HTTP keep-alive. To improve performance, modern web servers keep client connections alive to handle multiple requests instead of building up and tearing down connections for each item in a page. Keep-alive is tremendously beneficial to performance, but tends to keep unnecessary connections alive, too. By default, lighttpd allows 16 keep-alive requests per connection, allows idle sessions to remain alive for 5 seconds, and gives reads and writes 1 minute and 6 minutes to complete, respectively.
One reason that file descriptors get used up so quickly is HTTP keep-alive. To improve performance, modern web servers keep client connections alive to handle multiple requests instead of building up and tearing down connections for each item in a page. Keep-alive is tremendously beneficial to performance, but tends to keep unnecessary connections alive, too. By default, lighttpd allows 16 keep-alive requests per connection, allows idle sessions to remain alive for 5 seconds, and gives reads and writes 1 minute and 6 minutes to complete, respectively.
# Maximum number of request within a keep-alive session before the server terminates the connection, default = 16 (<code>server.max-keep-alive-requests</code>)
# Maximum number of seconds until an idling keep-alive connection is dropped, default = 5 (<code>server.max-keep-alive-idle</code>)
# Maximum number of seconds until a waiting, non keep-alive read times out and closes the connection, default = 60 (<code>server.max-read-idle</code>)
# Maximum number of seconds until a waiting write call times out and closes the connection, default = 360 (<code>server.max-write-idle</code>)


Although lighttpd has pretty aggressive defaults (especially compared to Apache), a period of heavy traffic and a few slow clients could see many unused connections sticking around. The server.max-keep-alive-idle setting default of 5 seconds can be reduced to as low as 2, if you assume your clients are reasonably quick about requesting data, but a value of 3 or 4 is probably realistic. You may want to increase the server.max-keep-alive-requests value from the default of 16, but you probably don’t need to. The server.max-read-idle and server.max-write-idle settings are tempting targets, but these situations are usually fairly rare so let’s not monkey with them.
Although lighttpd has pretty aggressive defaults (especially compared to Apache), a period of heavy traffic and a few slow clients could see many unused connections sticking around. The server.max-keep-alive-idle setting default of 5 seconds can be reduced to as low as 2, if you assume your clients are reasonably quick about requesting data, but a value of 3 or 4 is probably realistic. You may want to increase the server.max-keep-alive-requests value from the default of 16, but you probably don’t need to. The server.max-read-idle and server.max-write-idle settings are tempting targets, but these situations are usually fairly rare so let’s not monkey with them.
Line 266: Line 271:
=== Lighttpd_Advanced_security ===
=== Lighttpd_Advanced_security ===


See at [[Lighttpd_Advanced_security]] wikip page.
See at [[Lighttpd_Advanced_security]] wiki page.


=== Lighttpd and PHP with fpm ===
=== Lighttpd and PHP with fpm ===

Revision as of 22:49, 7 March 2020

lighttpd is a simple, standards-compliant, secure, and flexible web server.

Lighttpd Installation

Production environment only will handle need packages.. so no doc or manages allowed, Since kernel version 2.6, Linux has sys_epoll, a so-called edge-triggered polling mechanism which scales linearly with the number of connections. Due lighttpd runs over Alpine Linux only the mechanish we will use are only the "linux-sysepoll".

Lighttpd attempts to improve performance further by caching the output of the UNIX stat() command. It includes a basic (“simple”) cache which keeps the result of file system calls in memory for one second. But many Linux distributions include more advanced accelerators: FAM was the original, and a lighter-weight workalike called Gamin is now included by default in Ubuntu’s lighttpd install.

  1. run apk for need pacakges
  2. make the htdos public web root directories
  3. change default port to production one, http are used with 80
  4. use FAM stule (gamin) file alteration monitor, increases performance
  5. use linux event handler, increases performance due Alpine are linux only
  6. added the servide to the default runlevel, not to boot, because need networking activated
  7. started the web server service

apk add lighttpd gamin

mkdir -p /var/www/localhost/htdocs /var/log/lighttpd /var/lib/lighttpd

sed -i -r 's#\#.*server.port.*=.*#server.port          = 80#g' /etc/lighttpd/lighttpd.conf

sed -i -r 's#.*server.stat-cache-engine.*=.*# server.stat-cache-engine = "fam"#g' /etc/lighttpd/lighttpd.conf

sed -i -r 's#\#.*server.event-handler = "linux-sysepoll".*#server.event-handler = "linux-sysepoll"#g' /etc/lighttpd/lighttpd.conf

mkdir -p /var/lib/lighttpd

chown -R lighttpd:lighttpd /var/www/localhost/

chown -R lighttpd:lighttpd /var/lib/lighttpd

chown -R lighttpd:lighttpd /var/log/lighttpd

rc-update add lighttpd default

rc-service lighttpd restart

echo "it works" > /var/www/localhost/htdocs/index.html

For testing open a broser and go to http://<webserveripaddres> and you will see "it works". The "webserveripaddres" are the ip address of your setup/server machine.

There's a problem in Alpine linux, FAM (gamin) are activated as a lighttpd only service, that's make sense in dockers but in servers could be a problem if FAM (gamin) are also need for others services at the same time.

Controlling Lighttpd

Start lighttpd: After the installation lighttpd is not running. As we made in first section was started already but if you want to start lightttpd manually use:

rc-service lighttpd start

You will get a feedback about the status.

 * Caching service dependencies                                 [ ok ]
 * Starting lighttpd...                                         [ ok ]

Stop lighttpd: If you want to stop the web server use stop in the same way of previous command:

rc-service lighttpd stop

Restart lighttpd: After changing the configuration file lighttpd needs to be restarted.

rc-service lighttpd restart

Proper Runlevel: By default no services are added to start process, sysadmin must know what we want and what will services do, also other main reason are due in dockers there's no runlevels per se and Alpine linux are mostly used in dockers containers. You must added the servide only to the default runlevel, not to boot, because need networking activated

rc-update add lighttpd default

Lighttpd Configuration

If you just want to serve simple HTML pages lighttpd can be used out-of-box. No further configuration needed.

Due to the minimalism of alpine linux, unfortunately the lighttpd packaging is the worst ever seen, its configuration file makes it impossible to configure with only single line commands so the commands for quick configuration with cares of overwriting are very dedicated.

Status special page

Taking care of the status web server: those special pages are just minimal info of the running web server, are need to view from outside in a case of emergency, do not take the wrong approach of hide behind a filtered ip or filtered network, you must have access in all time in all the web to see problems. The creation of the directory in the htdocs main root web files are just to remember you so then can avoid hiring a staff that becomes indispensable, thus allowing to save costs in knowledge theft by technical staff.

  1. Enable the mod_status at the config files
  2. change path in the config file, we are using security by obfuscation
  3. restart the service to see changes at the browser

mkdir -p /var/www/localhost/htdocs/stats

sed -i -r 's#\#.*mod_status.*,.*#    "mod_status",#g' /etc/lighttpd/lighttpd.conf

sed -i -r 's#.*status.status-url.*=.*#status.status-url  = "/stats/server-status"#g' /etc/lighttpd/lighttpd.conf

sed -i -r 's#.*status.config-url.*=.*#status.config-url  = "/stats/server-config"#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

CGI bin directory support

By default packages assign a directory under localhost main domain, other linux uses a global cgi directory and aliasing.. the most profesional way, but think about it, this per domain configuration allows isolation:

  1. enable the mod_alias at the config file, due need of a specific path for cgi files into security
  2. create the directory due packager dont make any reference to that neither in the lighttpd-doc
  3. enable the config cgi file
  4. restart the service to see changes at the browser

mkdir -p /var/www/localhost/cgi-bin

sed -i -r 's#\#.*mod_alias.*,.*#    "mod_alias",#g' /etc/lighttpd/lighttpd.conf

sed -i -r 's#.*include "mod_cgi.conf".*#   include "mod_cgi.conf"#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

After that, all the files under the /var/www/localhost/cgi-bin directory will be showed as http://localhost/cgi-bin/ path

Plus this config file enables that all .cgi files are perl procesed.. that's wrong, but at the moment that are very specific, each development must document how to deploy property and only enables cgi in specific way.

Make special errors (404 or 500) pages for clients and visitors

This pages will be show to visitors when a page or path are not in the server, or when a internal error happened, this are to do not show a horrible message of development to visitors.. and just a nice message or "away from here" message:

  1. create the directory for put the html files to show when that errors happened in the way
  2. create the simple files for each message in the directory
  3. set the proper in the configuration file
  4. restart the service to see the changes at the browser (just request a non existing page and you will see it)

mkdir -p /var/www/localhost/errors

cat > /var/www/localhost/errors/status-404.html << EOF
<h1>The page that you requested are not yet here anymore, sorry was moved or updated, search or visit another one</h1>
EOF

cat > /var/www/localhost/errors/status-500.html << EOF
<h1>Please wait a moment, there's something happens and we are give support maintenance right now to resolve</h1>
EOF

cp /var/www/localhost/errors/status-404.html /var/www/localhost/errors/status-403.html

cp /var/www/localhost/errors/status-500.html /var/www/localhost/errors/status-501.html

cp /var/www/localhost/errors/status-500.html /var/www/localhost/errors/status-503.html

sed -i -r 's#.*server.errorfile-prefix.*#server.errorfile-prefix    = var.basedir + "/errors/status-"#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

Userdir public_html support

Lighttpd SSL support

The package as we said is made in a limited way, and only have vague references, in the configuration file for the SSL, only put two lines of configuration, and if you try to uncomment that lines, the service will not start, since there is no line for the openssl module and must be put manually.

Best way to do that are by external include files, Debian counterpart has a good mechanism that enables configuration files, we will made for SSL support in the same way.. all SSL related will be in a specific file.. but that file must be includen first thatn the rest of the configurations, but just after the modules loading, to make effect in https cases.

SSL : making self signed certificate

We need to created a sefl-signed certificate, so openssl are need in any case either if used a remote made certificate:

  1. install openssl
  2. create the self signed certificate
  3. set proper permissions
  4. create a SSL module configuration file for lighttpd
  5. activate the openssl module missing from config file
  6. activate the mod_redirect in case of global http to https redirections
  7. restart the service to see changes

apk add openssl

mkdir -p /etc/ssl/certs/

openssl req -x509 -days 1460 -nodes -newkey rsa:4096 \
   -subj "/C=VE/ST=Bolivar/L=Upata/O=VenenuX/OU=Systemas:hozYmartillo/CN=$(hostname -d)" \
   -keyout /etc/ssl/certs/$(hostname -d).pem -out /etc/ssl/certs/$(hostname -d).pem

chmod 640 /etc/ssl/certs/$(hostname -d).pem

cat > /etc/lighttpd/mod_ssl.conf << EOF
server.modules += ("mod_openssl")
\$SERVER["socket"] == "0.0.0.0:443" {
	ssl.engine  = "enable"
	ssl.pemfile = "/etc/ssl/certs/$(hostname -d).pem"
	ssl.cipher-list = "ECDHE-RSA-AES256-SHA384:AES256-SHA256:RC4:HIGH:!MD5:!aNULL:!EDH:!AESGCM"
	ssl.honor-cipher-order = "enable"
}
\$HTTP["scheme"] == "http" {
    \$HTTP["host"] =~ ".*" {
        url.redirect = (".*" => "https://%0\$0")
    }
}
EOF

sed -i -r 's#\#.*mod_redirect.*,.*#    "mod_redirect",#g' /etc/lighttpd/lighttpd.conf

checkssl="";checkssl=$(grep 'include "mod_ssl.conf' /etc/lighttpd/lighttpd.conf);[[ "$checkssl" != "" ]] && echo listo || sed -i -r 's#.*include "mime-types.conf".*#include "mime-types.conf"\ninclude "mod_ssl.conf"#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

For deploy usage of Lets Encrypt without chain-tools (just add water) read Production Lets Encrypt: dehydrated.

Lighttpd advanced

Lighttpd has pretty good default settings, but a few might be tweaked if we need to respond to higher server loads. The more important area of tuning is simply enabling the advanced features of the recents Alpine Linux kernel we are using: Enable sys_epoll, sendfile and disable atime updates.

Lighttpd tunning for agressive load

Use noatime and dont care of cache ONLY FOR MANY STATIC FILES: although it’s not included in the official lighttpd performance document, is not updating the "atime" parameter on served pages. This is a bit of a religious issue among some UNIX administrators becose the monitoring cases, BUT we almost always mounting the entire filesystem with “noatime”, but more granular approach offered by lighttpd:


checkset="";checkset=$(grep 'noatime' /etc/lighttpd/lighttpd.conf);[[ "$checkset" != "" ]] && echo listo || sed -i -r 's#server settings.*#server settings"\nserver.use-noatime = "enable"\n#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

Mechanics: Polling and Sending: By default in this document we set the triggered polling mechanism server.event-handler to linux-sysepoll. But how to actually read and write data from the disk to the network in Linux includes a more advanced call, sendfile, which can move data around without copying it into memory. We can enable this by setting "server.network-backend" to "linux-sendfile", which ought to improve performance for larger (multi-megabyte) files without impacting smaller ones.


checkset="";checkset=$(grep 'network-backend' /etc/lighttpd/lighttpd.conf);[[ "$checkset" != "" ]] && echo listo || sed -i -r 's#server settings.*#server settings"\nserver.network-backend = "linux-sendfile"\n#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart

More connections, More File Descriptors

This must be used with caution, everything is a file to a UNIX operating system. Well, every time a visitor accesses a page, lighttpd uses three file descriptors: An IP socket to the client, a fastCGI process socket, and a filehandle for the document accessed. Lighttpd stops accepting new connections when 90% of the available sockets are in use, restarting again when usage has fallen to 80%. With the default setting of 1024 file descriptors, lighttpd can handle a maximum of 307 connections. If this number are exceded file descriptor must be increrased then. This are a delicate tune due must be check your default with cat /proc/sys/fs/file-max and make sure it’s over 10,000:


checkset="";checkset=$(grep 'max-fds' /etc/lighttpd/lighttpd.conf);[[ "$checkset" != "" ]] && echo listo || sed -i -r 's#server settings.*#server settings\nserver.max-fds = 2048\n#g' /etc/lighttpd/lighttpd.conf

rc-service lighttpd restart


HTTP Keep-Alive for aggressive load

One reason that file descriptors get used up so quickly is HTTP keep-alive. To improve performance, modern web servers keep client connections alive to handle multiple requests instead of building up and tearing down connections for each item in a page. Keep-alive is tremendously beneficial to performance, but tends to keep unnecessary connections alive, too. By default, lighttpd allows 16 keep-alive requests per connection, allows idle sessions to remain alive for 5 seconds, and gives reads and writes 1 minute and 6 minutes to complete, respectively.

  1. Maximum number of request within a keep-alive session before the server terminates the connection, default = 16 (server.max-keep-alive-requests)
  2. Maximum number of seconds until an idling keep-alive connection is dropped, default = 5 (server.max-keep-alive-idle)
  3. Maximum number of seconds until a waiting, non keep-alive read times out and closes the connection, default = 60 (server.max-read-idle)
  4. Maximum number of seconds until a waiting write call times out and closes the connection, default = 360 (server.max-write-idle)

Although lighttpd has pretty aggressive defaults (especially compared to Apache), a period of heavy traffic and a few slow clients could see many unused connections sticking around. The server.max-keep-alive-idle setting default of 5 seconds can be reduced to as low as 2, if you assume your clients are reasonably quick about requesting data, but a value of 3 or 4 is probably realistic. You may want to increase the server.max-keep-alive-requests value from the default of 16, but you probably don’t need to. The server.max-read-idle and server.max-write-idle settings are tempting targets, but these situations are usually fairly rare so let’s not monkey with them.

Lighttpd_Advanced_security

See at Lighttpd_Advanced_security wiki page.

Lighttpd and PHP with fpm

In production web, LAMP means Linux + Apache + Mysql + Php installed and integrated, but today the "A" of apache are more used as Nginx or Lighttpd, and the "M" of MySQL are more used as Mariadb, the LAMP focused documents are:

See Also