Daemons and Paths

[Davertron]

So I've written a daemon using the 'daemons' gem. I have essentially
3 pieces:

daemon_controller => Controls startup/shutdown of daemon (in the style
of the first example of the daemons documentation)

daemonized_script => This script is the main program; it runs
continuously in the background looking for work to do, and when it
finds work, it forks itself and uses 'exec' to run a worker

worker => processes the job given to it by the daemonized_script and
exits

The problem I'm having is that my worker program expects to be able to
'require' a few other classes I've written. It expects them to be in
the same directory as it (or in a lib directory one dir up,
whatever). The obvious issue here is that when the daemon_controller
executes the daemonized_script as a daemon, it changes the directory
to '/' and executes from there. I temporarily fixed this issue by
putting in absolute paths in my source code to test and verify that
everything else works as expected, which it does, so my question is
this; how do I get a dynamic path set so that the worker process
doesn't error when it can't find the files it needs? I can't use
absolute paths because my program won't always be installed to the
same directory, and anyway that's not a very flexible solution. I
should also mention that all of these files live in the same directory
structure, so relative paths should work fine. In other words, I'm
not putting daemon_controller somewhere else in the filesystem and
then trying to call my other files from there; they'll all be either
in the same directory, or in close proximity to each other so that I
only have change a directory up or down to get to the other files.

Any response will be much appreciated.

 

[Michael Linfield]

I don't know if this is exactly what your looking for, but how about a
call to locate?

begin
#worker code here

rescue
path = %x[locate <filename>].chomp! #the filename would be whatever the
daemon is
#looking for - you take the error
code from
#the daemon and pull the filename
out of it

#Here we are going to take the path output
#then remove the last array element it's
#split by, aka the filename, so that u will
#have the directory

directory = path.split("/")
directory.pop
directory.join("/")
Dir.chdir(directory)

retry #retries the begin block

end

 

[Michael Linfield]

I don't know if this is exactly what your looking for, but how about a
call to locate?

begin
#worker code here

rescue
path = %x[locate <filename>].chomp!

#the filename would be whatever the daemon is
#looking for - you take the error code from
#the daemon and pull the filename out of it

#Here we are going to take the path output
#then remove the last array element it's
#split by, aka the filename, so that u will
#have the directory

directory = path.split("/")
directory.pop
directory.join("/")
Dir.chdir(directory)

retry #retries the begin block

end

My apologies, correct the line that says directory.join("/") to the
following:

directory = directory.join("/")

Regards,

- Mac

 

[Davertron]

I don't know if this is exactly what your looking for, but how about a
call to locate?

begin
#worker code here

rescue
path = %x[locate <filename>].chomp!
#the filename would be whatever the daemon is
#looking for - you take the error code from
#the daemon and pull the filename out of it

#Here we are going to take the path output
#then remove the last array element it's
#split by, aka the filename, so that u will
#have the directory

directory = path.split("/")
directory.pop
directory.join("/")
Dir.chdir(directory)

retry #retries the begin block

end

My apologies, correct the line that says directory.join("/") to the
following:

directory = directory.join("/")

Regards,

- Mac

Thanks Michael. You're right, it's not exactly what I'm looking for,
but it would probably work. I was looking more towards modifying my
RUBYPATH environment variable to get this work; something that doesn't
require putting in a bunch of special-case code inside of the source
itself, especially above all of my "require" statements. I was
thinking of writing a "locate_and_require" statement with your
suggestions above if there really is no "right" way to do this. I've
never written daemons before, so I don't really know how people
generally handle this, and perhaps my expectations are out-of-whack.
Anyway, I did find code that shows modifying of the $: variable inside
a script to add/change your ruby path; however, it doesn't look like
the changes survive daemonization. I'll keep looking around and post
how I solved this once I get there.