On Nov 12, 2007 8:53 AM, chago <nabble@visualxs.com> wrote:

I know nearly nothing about linux, although I am running Ubuntu on my mac
TiBook (ppc).

I installed alien which can install rpms and convert between many install
packages. I used it to generate a pre/post scripts and unpack the rpm into a
local directory. I can also use alien to install the rpm. Other than some
inappropriate (for ubuntu) directories, it seems like it would install fine.

I really don't have the skill to write an install script. I was baffled by a
lot of the /etc/init.d/apacheds file, especially the recursive launch of the
script when run-as does not match current user. I'm not even sure what
 
This was inherited from the Tanuki wrapper daemon and is basically a mechanism to launch the java program as a different user.  This is a necessary hack since Java doesn't do setuid   I didn't like a lot of that either, but the Tanuki wrapper works really well and has a lot of potential.  Maybe we can make some improvements or suggestions to them.


directories are preferred when installing the different sections of the
code. For example, the current codebase works with these separate
directories...

/etc/init.d - Used for apacheds start/stop script (same for ubuntu)
/etc/sysconfig - Used for specifying a couple directory paths (don't know
where ubuntu would want these. Seems like a waste of a folder for just
storing some "conf" data. I suppose they exist because they are conf data
for the server proper and not the server instance.)

Exactly...  this is just to have a place to set global env data before the rest of the script runs so that the admin doesn't have to edit the init script.  This lets these customizations survive upgrades.


/var/lib/apacheds - Used for configuration and runtime data (partitions).
Also holds subdirectories for different configurations of apacheds. Comes
with a 'default' config. (Maybe /usr/local/apacheds/configs?)

This directory is specified by the FHS specs ( http://www.pathname.com/fhs/ ) and is used by many other programs that have data storage needs (MySQL, Postgres, etc.)


/var/run - Holds the pid apacheds is running on. (/var/lock in ubuntu)
/opt/apacheds-1.5.1 - Holds the implementation of the server, including the
bin directory for launching. (not sure where this should go, maybe
/usr/local/apacheds-1.5.1?)

This is again specified by FHS, but as long as the location is reflected in the init script, it can go anywhere.
 


So, you can see I don't know much about it, but I may know enough to get it
up and running although in a very manual manner. Assuming the init.d script
works on Ubuntu. The init.d script wants the name of an instance when
starting/stopping. How does it find this name parameter when the system is
starting up the server? It doesn't seem to default to anything...

Yeah, this was an oversight and I think we are going to have to figure out a better way to do this in the next release.  There are basically 2 options: we create a config that says which instances get started by default (so starting stopping the service starts/stops all named instances ), or we create a new init for each instance.  I have seen it done both ways so suggestions are welcome.
 


FWIW, the i386 rpm seems to work OK on my PPC architecture.
Yes, with the Java installers, the platform id is largely irrelevant since Java has to be there seperately.  So I guess we could just make copies for the other platforms in case we ever do a yum repo or any of the other distro repos.

Thanks for the discussion.  Like Alex said, any help in this area is welcome.

Chris



Alex Karasulu wrote:
>
> You're right.  It's lame that we have all these GUI installers and no
> tarball. Someone has to
> scratch this itch.  Wanna give it a try?
>

--
View this message in context: http://www.nabble.com/-ApacheDS--In-preparation-for-1.0.1-release%3A-candidates-available-tf3268468.html#a13708105
Sent from the Apache Directory Project mailing list archive at Nabble.com.