Wishlist request: SSH remote administration added to API

[Aaron G. <agr...@uw...>]

First of all, thank you very much for webmin.  It is a very valuable
tool.  I think I have come up with a way for it to be even more of a
help to administrators.  My suggestion is this:

Currently, many plugins require that webmin be loaded on the same box as
the service being administered.  This makes webmin somewhat of an
administrative burden itself on a medium-to-large network. 
Although it offsets this by making administration easier, it is not
really necessary to do things this way.  There is already an
administrative service that ships with almost all unix-type systems, and
can be loaded on most systems that do not ship it.  I am, of course,
referring to ssh.

Every function (AFAICT, please let me know if I am mistaken here) that
is currently performed by webmin can be performed remotely via ssh.  In
addition, many things that webmin cannot do right now (i.e. Windows
remote administration) can be done via ssh, since it absolves the webmin
server of necessarily supporting the commands being performed.  Only the
plugin has to handle the commands, and be able to locate and deal with
them remotely.

Proposed API additions:

ssh_exec - run commands via ssh and return their output for
error-handling
ssh_get_file - download a file either via ssh and cat, or scp
ssh_put_file - the reverse
ssh_open - a wrapper for open() which uses the above commands to
retrieve a file before opening, then put it back when finished.  These
sorts of commands would be important for porting, as the external plugin
authors are probably not going to want to put all the logic in, and it
should be able to be handled centrally in the API for most cases.

And whatever others I am forgetting that are important for this to work
right.  How's that for open-ended? ;)

I really think that by further relieving administrators' burdens wrt
webmin cluster maintenance and opening the way into non-unix systems for
webmin plugins this would make webmin even more of a must-have tool than
it already is.

Of course, you are probably thinking "well, so why don't you do it
then?"  The answer is that I lack the skill to do this myself.  My perl
skills are limited, although given sufficient direction I would be
willing to help with, say, updating the plugins once the API changes
were made.  That shouldn't be rocket science, probably just seriously
boring except in the corner cases.

I hope that this is not a huge task, but I think that it could be a huge
benefit to this project.  You know the code better than I, as I have
only reviewed it briefly.  What do you think?


Thanks much,
-Aaron Grewell