requirements to start a process under process management

From: Avery Payne <avery.p.payne_at_gmail.com>
Date: Fri, 10 Oct 2014 11:40:48 -0700

Digging through how the template currently works, it looks like there are
several assumptions being made "up front" about what is required to launch
a program. The more I dig, the more I wonder how much of this could be
boiled down and condensed.

+ path of the program
+ path of any required libraries (if dynamically linked)
+ if a multi-part program, presence of all the other parts needed to start
or function
+ program-specific options as parameters (i.e. --some-option)
+ environment variables
+ a launcher (typically shell or a purpose-built program like execline)
+ a run directory to store settings, pid files (ick), sockets, etc.
+ configuration file(s)
+ presence of specific files, such as /dev entries, socket files, etc that
are not set by the program but assumed to be there
+ presence of specific directories (example: postgresql wants a data
directory for storing its tables)

Please point out anything that is missing. It helps to have a full picture
of what the expectations are, especially when they are "passed by oral
tradition" instead of written down somewhere. Thanks for any input in
advance.
Received on Fri Oct 10 2014 - 18:40:48 UTC