speedy(1) a persistent Perl interpreter


speedy [ <perl options> ] [ -- <speedy options> ] [ <filename> ]


speedy, short for SpeedyCGI, is a way to run perl scripts persistently, which can make them run much more quickly. The most common way to make a script run persistently is by changing the interpreter line at the top of the script from:



After the script is initially run, instead of exiting, the perl interpreter is kept running. During subsequent runs, this interpreter is used to handle new executions instead of starting a new perl interpreter each time. A very fast frontend program, written in C, is executed for each request. This fast frontend then contacts the persistent Perl process, which is usually already running, to do the work and return the results.

Each perl script runs in its own Unix process, so one perl script can't interfere with another. Command line options can also be used to deal with programs that have memory leaks or other problems that might keep them from otherwise running persistently.

Although SpeedyCGI is especially useful for CGI scripts, it can be used to keep any perl script running persistently. See CGI::SpeedyCGI(3.pm) for a complete description of SpeedyCGI, including further details on CGI execution and Apache issues. This manual page is based on that, but concentrates on running speedy from the command line.


The speedy command line is the same as for regular perl, with the exception that SpeedyCGI specific options can be passed in after a "--".

For example the line:

#!/usr/bin/speedy -w -- -t300

at the top of your script will set the perl option `-w' and will pass the `-t' option to SpeedyCGI, setting the Timeout value to 300 seconds.

The options can also be set at run-time from the perl script using the CGI::SpeedyCGI module.


See CGI::SpeedyCGI(3.pm) for a complete description of the options.
BackendProg : Path to the speedy backend program. (Default: /usr/bin/speedy_backend)
BufsizGet : Use <number> bytes as the maximum size for the buffer that receives data from the perl backend. (Default: 131072)
BufsizPost : Use <number> bytes as the maximum size for the buffer that sends data to the perl backend. (Default: 131072)
Group : Allow a single perl interpreter to run multiple scripts. See CGI::SpeedyCGI(3.pm) for details. (Default: `none')
MaxBackends : If non-zero, limits the number of speedy backends running for this perl script to <number>. (Default: 0)
MaxRuns : Once the perl interpreter has run <number> times, re-exec the backend process. Zero indicates no maximum. This option is useful for processes that tend to consume resources over time. (Default: 500)
Timeout : If no new requests have been received after <number> seconds, exit the persistent perl interpreter. Zero indicates no timeout. (Default: 3600)
TmpBase : Use the given prefix for creating temporary files. This must be a filename prefix, not a directory name. (Default: `/tmp/speedy')
Version : Print the SpeedyCGI version and exit.


Environment variables can also be used to pass in options. This can only be done before the initial execution, not from within the script itself. The name of the environment variable is always SPEEDY_ followed by the option name in upper-case. For example to set the speedy Timeout option, use the environment variable named SPEEDY_TIMEOUT.


/tmp/speedy* A unix socket used to connect to the backend process. See speedy_backend(1) for more information.


 Sam Horrocks
 [email protected]


This manual page was created by Niko Tyni <[email protected]> for Debian GNU/Linux, because the original program does not have one. It is based on the original and more complete CGI::SpeedyCGI(3pm) manual page.


There are command-line parsing incompatibilities with the real Perl. These aren't very easy to fix, as even the perlrun manpage isn't quite accurate on which parameters can be separated (like '-I') and which can't (like '-C'). speedy doesn't allow any of them to be separated. It considers the first option without a leading dash as the script filename.