Re: Standing orders queue

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

 



On Thu, 2008-10-09 at 10:32 -0400, Michael DeHaan wrote:
> seth vidal wrote:
> > On Mon, 2008-10-06 at 17:39 -0400, Michael DeHaan wrote:
> >   
> >> Michael DeHaan wrote:
> >>     
> >>> It would be nice to have a way to issue offline orders to minions so 
> >>> they could check in when they come online (via XMLRPC) and then see if 
> >>> they have any orders.
> >>>
> >>> This is the other mode of communication we would like to support in 
> >>> order to enable Func to replace OSAD in Spacewalk.
> >>>
> >>> For example, "this node should yum update the next time I see it" is 
> >>> different from "this node should update now if it's online".
> >>>
> >>> Anyone have any thoughts on implementations?
> >>>
> >>> --Michael
> >>>
> >>> _______________________________________________
> >>> Func-list mailing list
> >>> Func-list@xxxxxxxxxx
> >>> https://www.redhat.com/mailman/listinfo/func-list
> >>>       
> >> To elaborate, basically with what we have here and also the new cobbler 
> >> "config file templating" feature, we have the makings of a much easier 
> >> to understand config management tool, all in Python, and all very easy 
> >> to understand.   This could be the beginings of "RRS" (remote rocket 
> >> surgery) built on Cobbler+Func.   (See also recent Cobbler integration).
> >>
> >> We just need a way to address offline nodes, basically.
> >>
> >>     
> >
> > Offline commands are fine if you don't expect some sort of response from
> > the command, right?
> >
> > We'd need something more or less like this:
> >
> > for each host
> >    if ping host
> >       do command
> >    else
> >       store command and retry
> >
> > the retry trick only works well if we have another daemon running that
> > stores the commands and the responses, I think, I guess we could do it
> > with a regularly scheduled cron job to look for unrun jobs but....
> >
> > -sv
> >
> >   
> 
> I was originally thinking what if the Func minions just pull from the 
> queue?   The only problem with that is we don't have a programatic easy 
> way to look at all the results together as we do with async today.   
> That would probably be harder and maybe not as useful.
> 
> It would be nice if it worked using the existing async style API's.

What's the model we're targeting with regard to interactivity? We don't
really want highly interactive interfaces but do any of our modules
right now do anything more than just issue commands and get result
codes/returned data?

We could have the daemon store both commands and results and wait for
them to be viewed.

something like:

/var/spool/func-queue/
                      hostname/commands/0/1/2/012345
                              /results/0/1/2/012345

Or at this point are we better off just using xmpp?

-sv



_______________________________________________
Func-list mailing list
Func-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/func-list

[Index of Archives]     [Fedora Users]     [Linux Networking]     [Fedora Legacy List]     [Fedora Desktop]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite News]     [KDE Users]

  Powered by Linux