[Linux-ha-dev] LRM / CRM: RA Metadata
Andrew
lists at beekhof.homeip.net
Tue Jun 29 09:26:40 MDT 2004
On Jun 29, 2004, at 4:37 PM, Lars Marowsky-Bree wrote:
> On 2004-06-29T15:32:01,
> Andrew <lists at beekhof.homeip.net> said:
>
>> Ok, I think see where you're going now.
>>
>> I had been thinking of 2 configured layers (Global and per resource)
>> and an automated layer (per RA from the LRM). But if understand you
>> right, you want the per resource and per RA but both configured rather
>> than generated.
>>
>> Correct?
>
> Yes. But I'm not sure whether it's the best way ;-)
>
> If we had all the data in the CIB, it would sure be easier, alas the
> message size might become a problem. However, maybe it's better to
> address that instead of doctoring around it.
>
> The layers I'd forsee though, however they are stored, from top to
> bottom:
>
> - Global defaults for a resource class, configurable
> (to have a default set of parameters for LSB init scripts for
> example)
> - defaults for a resource type (defaults supplied by RA)
> - defaults for a resource type (configurable)
> - defaults for a resource type + version (defaults supplied by RA)
> - defaults for a resource type + version (configurable)
>
> - Resource configuration in the CIB
> - Resource configuration in the CIB but with a more specific node
> filter
>
> Some of the layers are optional, of course; the only one really needed
> is the "resource" entity in the CIB.
doooode! do you ever want this finished? :)
Do you seriously want all those? I'm not convinced about the +version
ones. When would they get cleaned out? Is going from 1.0.2.1 to
1.0.2.2 significant? I am thinking that we'd be better off taking the
IPaddr approach and having IPaddr, IPaddr2, ... when there is a
significant change in options/symantics/implementation/whatever.
With the last one, is that the conditional resource parameters?
Hmmm... here's one for you, what if you want to use IPaddr2 on all your
linux nodes but IPaddr on the Solaris box (since it may not support
IPaddr2)?
>
>
> Sincerely,
> Lars Marowsky-Brée <lmb at suse.de>
>
> --
> High Availability & Clustering \ ever tried. ever failed. no
> matter.
> SUSE Labs, Research and Development | try again. fail again. fail
> better.
> SUSE LINUX AG - A Novell company \ -- Samuel Beckett
>
> _______________________________________________________
> Linux-HA-Dev: Linux-HA-Dev at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev
> Home Page: http://linux-ha.org/
More information about the Linux-HA-Dev
mailing list