David,
I agree that the problem is a context or scope change. But do you
require that every resource that is just coming up (hence getting a new
context) or any resource that goes through a power cycle to issue a new
event in addition to the other events that are related to the state
change that it will send and will signify a context change? I am not
sure if we want the additional event here... but I am flexible on this.
Cheers,
H.
--
David E Cox wrote:
I think the real point is not
whether
the resource is dead or down or inactive. The point is whether the
resource has lost the "context" it used to ensure that event
identifiers are unique. That context could have been the time, a
counter, or some other modifier for uniqueness.
What if we put the requirement that
as soon as a resource establishes a new context for uniqueness of event
identifiers, it either sends an event stating so (preferred), or puts
something
unique in the first event it sends, so that the manager doesn't have to
"guess".
Regards,
David E Cox
What I mean by "dead" is that the scope changes
only if a resource becomes completely inactive (e.g. a power cycle).
If it is just dormant, then after it becomes active, there is no change
of scope. A consumer (manager) should be able to tell the difference
when
something is just dormant or gone.
H.
---
Sedukhin, Igor S wrote:
oh, so you want it dead :):).... say it pretends
really
well do be dead... there is no way to tell from the consumer's
prespective
and that is what matters: activity of the implementation of a
manageable
resource from the POV of the manageability consumers.
-----Original Message-----
From: Homayoun Pourheidari [mailto:homayoun@bea.com]
Sent: Mon 11/15/2004 5:49 PM
To: Sedukhin, Igor S
Cc: wsdm@lists.oasis-open.org
Subject: Re: [wsdm] managed domains and unique event identifiers
"managed resource" should be changed to "manageable
resource" as defined
in the MUWS spec.
"go up and down" was the only way I could think of to describe
that a
resource may be alive or not. I am open to suggestions...
I used the term "managed domain" since there seemed to some specific
interest in this term. Nonetheless I am also more comfortable with
the
term "managed scope". The term "management scope"
gives me the
impression that the scope of the manager is also part of it -- which is
not.
"activity scope" could mean that something is active and then
dormant
for a while (not necessarily dead). Hence, I am not comfortable with
this term.
Cheers,
H.
--
Sedukhin, Igor S wrote:
>1. "managed resource" needs to be defined. It is not defined
in MUWS
>now.
>2. "go up and down" needs to be defined. It is not clear
what that is.
>We may all be thinking of different things reading it.
>4. "managed domain" or "management domain" or "manageability
domain"?
>Which would be the right term here? This seems more like "management
>scope" than a "domain".
>5. Creation of "managed domain" is a bit confusing, IMO.
>
>I suggest that we simply say "Event identifiers are unique for
only the
>activity scope of the implementation of the manageable resource
which
>observers the events. That is, if activities of an implementation
of
a
>manageable resource are interrupted, such as due to a power cycle,
the
>activity scope is terminated. After that, event ideftifiers would be
>unique to another activity scope of the same implementation of a
>manageable resource.
>
>
>
>-- Igor Sedukhin .. (igor.sedukhin@ca.com)
>-- (631) 342-4325 .. 1 CA Plaza, Islandia, NY 11749
>
>
>-----Original Message-----
>From: Homayoun Pourheidari [mailto:homayoun@bea.com]
>Sent: Monday, November 15, 2004 3:30 PM
>To: wsdm@lists.oasis-open.org
>Subject: [wsdm] managed domains and unique event identifiers
>
>Hi,
>
>Below is a porposed text to describe managed domains and the unique
>event identifier requirements for wsdm events.
>
>Cheers,
>H.
>--
>
>A managed resource may go up and down many times during its life
cycle.
>
>Every time that it is alive or up, it has an associated scope that
>includes all state and property values of the resource among other
>manageability information. We call this scope the managed domain
of a
>resource.
>
>For all events associated with the scope of a resource, all
>notifications that describe the events and are created to report
them,
>must have unique identifiers. The identifiers are not required
but may
>also be unique globally across the collective managed domains of
all
the
>resources that a manager is managing.
>
>When a resource is forced to restart its scope (e.g. goes down and
comes
>back up) a new managed domain for that resource is also created.
WSDM
>does not require that notification ids produced in the latter
managed
>domain to be unique across the current and the former managed
domains
of
>the resource. However, more capable managers may provide ways
to
>preserve some continuity of scope between various instantiations of
a
>resource's managed domains as a way of providing a longer
perspective
on
>the life cycle of a managed resource. Unique identifiers
for
>notifications across all of the managed domains of a managed
resource
>may be one such candidate for continuity.
>
>
>
>
>
>
>[attachment "homayoun.vcf" deleted by David E Cox/Raleigh/IBM]
To unsubscribe from this mailing list (and
be removed
from the roster of the OASIS TC), go to
http://www.oasis-open.org/apps/org/workgroup/wsdm/members/leave_workgroup.php.
|