I am currently writing a fencing agent to handle the fencing of virtual machines that are running on top of XenServer hosts and I had a couple of questions related to making a Red Hat friendly version. Let me know if this is not the right place to be asking.
I have written a first version located here
Now this works fine via command line options, however I have found the following link which says that the input will be taken as name/value pairs from standard in. The link I found to this doc is around 3 years old and I was wondering if it is still valid before making any changes? Looking at fenced this does still seem to be the case.
I was also wondering if there is any benefit to using the fencing.py library over creating a stand-alone script? Is this the suggested method of implementing a new fencing agent for Red Hat? I have actually started on a version that uses this library, is there someone on this list I can talk to about making changes to it (to take a URL instead of a ip address, and to include a UUID parameter in the options assoc array). So far I've only made a couple of changes to it, but I think a few more are coming as it isn't quite happening for me just yet.
Lastly, I am using XenAPI.py (which is released by Citrix under the LGPL). This is more of a licensing general question, but am I ok to just include this file in my release? Reading the LGPL seems to suggest this is ok as long as the copyright notice stays, but I thought I'd ask the question just in case... I'd hate to go to the trouble of finishing this only to find out that it can't be released...
Thanks for your time :)