Project

General

Profile

Running Lighttpd under daemontools/supervise

Lighttpd is well suited to run monitored by the "supervise" program from Daniel J. Bernstein's daemontools package.

To set it up, you must first create a standard service directory structure. In the following example the service directory from where the process starts will be /srv/lighttpd-main/. It consists of the following directory and files:

/srv/lighttpd-main/           # the service directory
/srv/lighttpd-main/run        # the script that starts lighttpd
/srv/lighttpd-main/root/      # lighttpd configuration files go here
/srv/lighttpd-main/log/       # the service directory for the logger
/srv/lighttpd-main/log/run    # the script that starts the logger
/srv/lighttpd-main/log/main/  # log files go here

If you're using Gentoo, use /var/services instead of /srv

Note: the "./root/" directory is optional and could be called anything, and we use it here simply to have somewhere to put the configuration file_s_ (we chop up the configuration into several files and use the include statement).

File ./run

#! /bin/sh

exec 2>&1
exec lighttpd -D -f ./root/lighttpd.conf

File ./root/lighttpd.conf

Apart from your specific settings you probably want to include the following definitions:

server.pid-file = "./root/lighttpd.pid" 
#server.errorlog must be NOT be set

File ./log/run

#! /bin/sh

umask 0027
exec setuidgid nobody multilog ./main

Directory ./log/main/

The user that is specified to run multilog (nobody in the above example) must have write access to this directory:

chown nobody ./log/main/


After the configuration file is in place, just place a symlink into the svscan service directory (/service in DJB's examples):

ln -s /srv/lighttpd-main /service/

(Please visit Wayne Marshalls overview in case you're lost at this point).

That's all you need to get started.

See also LighttpdUnderSuperviseExampleConfiguration.

Runit

See also HowToRunLightyWithRunit (for graceful restart wrapper)