[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha-dev
Subject:    [Linux-ha-dev] Re: [Bug 105] Convert to OCF RA
From:       Lars Marowsky-Bree <lmb () suse ! de>
Date:       2005-02-19 8:45:30
Message-ID: 20050219084530.GH7016 () marowsky-bree ! de
[Download RAW message or body]

On 2005-02-19T12:08:47, Horms <horms@verge.net.au> wrote:

> > ------- Additional Comments From lmb@suse.de  2005-02-18 04:31 -------
> > I correct myself, some have been changed so that they take several
> > parameters, but not the IPaddr one. Maybe you'll find time to go
> > through the agents once more and clean them up.
> > 
> > This must be done before 2.0, because we can't change the resource
> > parameters in a released version!
> 
> Hi Lars,
> 
> could you plese point be towards some more information on what should
> be happening here.

Hi Horms, thanks for your interest! But this bug didn't really concern
you, you just got the mail because you are on one depending on it.

However, it appears to be one of those bugs where you have exactly the
right skill set for it, so why don't I explain it anyway ;-)

As part of converting the heartbeat resource scripts to OCF Resource
Agents for the 2.0 series, the parameters the scripts take ought to be
meaningful changed from the list (with often implied special meaning in
what is really a single parameter in the script itself, like how IPaddr
splits up the ip/netmask/broadcast etc internal) to a set of key/value
pairs, and the other metadata filled in sensibly.

And the script should be converted to comply to all OCF RA
functionality, and redundant options (like, the "info" one) should be
removed. (In particular as it often gives wrong information.)

Further, this is a good time to overhaul the scripts and clean up the
shell script syntax which has accumulated over the years, and move stuff
to a common library, like common logging functions for the scripts and
maybe more. (I've got to say that I think in this case the options
provided by the FailSafe script library 2 weren't too bad.)

I've tried to do that with the IPaddr2.in script as an example, but I
rewrote that before the LRM/CRM could really drive it, so I'm not sure
if it works at all, it was more of a proof of concept.  And of course,
the common library would evolve as more scripts were converted.

The IPaddr2.in script also tried to take advantage of the multiple
incarnations for the cluster IP; again, multiple incarnations weren't
yet done when I wrote that, so the exact naming of the variables
probably needs overhaul.

drbd.in is also a proof-of-concept, which doesn't yet match the not-yet
written master/slave code in the CRM, nor the exact definitions of how
drbd 0.8 works. However, the master/slave mode might give people an idea
of how the ldirectord, replicated iptables or other replicated services
might take advantage of the new support for master/slave style
resources.

If you wanted to step in here and help out, that would surely be great!
Just add yourself to the bugzilla and give the names of the scripts you
want to take & overhaul, so that noone does double work.


We shall be obliged,
	Lars

-- 
High Availability & Clustering
SUSE Labs, Research and Development
SUSE LINUX Products GmbH - A Novell Business

_______________________________________________________
Linux-HA-Dev: Linux-HA-Dev@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
Home Page: http://linux-ha.org/
[prev in list] [next in list] [prev in thread] [next in thread]