[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Linux-cluster] Clarification on the FenceAgentAPI

On 10-03-08 04:03 AM, Marek Grac wrote:

On 03/08/2010 06:13 AM, Digimer wrote:
There seems to be some confusion in the API
(http://sources.redhat.com/cluster/wiki/FenceAgentAPI) and whether
'option' or 'action' is the proper argument used to tell the fence
agent what to do. Currently I check for both, but would like some
clarity on this.

Option 'action' (getopt: -o, --action) is preferred and should be
supported in every fence agent.

More confusing though is this:

Per email conversation, monitor is also a required operations for new
agents and if it is single port hw then it will call 'status', if it
is multi port hw then it will call 'list'. In both cases it should not
print output

The API document doesn't talk about the 'monitor' or 'list' arguments.
Can someone clarify what these actions are support to request from the

'monitor' - test if fence agent is working as expected, you can login
and do a simple operation.

'list' - should list all available devices/VM/.. proper format is csv:
#id, #alias
* more columns can be added but they are not standardized and used yet
* lines are not ordered in any way


Thanks for the reply, Marek!

  I've setup my 'list' action to display like so:

digimer lework:~$ fence_na -a ariel.alteeve.com -n 2 -l ariel -p gr0tt0 -o list

  Is the compliant output then? The format is:

Or would it be better to treat anything after the '<node>,<node_alias>' as reserved and not print the status info?


PS - Thanks for the clarification on the 'option' vs. 'action', I've updated my agent and docs.

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]