(I'm sorry for the long delay between Ira's message and my response--my
wife went into labor about two hours after Ira's message, and then I was on
leave for a while.)
Thanks for the reply, Ira. My responses are below prefaced by DC>.
Dennis
"McDonald, Ira" <imcdonald at sharplabs.com> on 04/03/2000 10:40:59 AM
To: Dennis Carney/Boulder/IBM at IBMUS, "McDonald, Ira"
<imcdonald at sharplabs.com>, "'ipp at pwg.org'" <ipp at pwg.org>
cc: Harry Lewis/Boulder/IBM at IBMUS
Subject: RE: IPP> FW: Slides for Job Mon MIB trap extensions for PWG-C
Hi Dennis, Monday (3 April 2000)
(I've copied the IPP list on this reply, because the thread is
probably of general interest)
Actually these three attributes in are RECOMMENDED (SHOULD) in
the SNMP trap bindings and not merely OPTIONAL (MAY).
DC> Good! I missed that.
They are only OPTIONAL (MAY) in the IPP Event Notification spec
(8 March 2000) in the table on page 42.
For the SNMP trap bindings, I couldn't reasonably make these
attributes MANDATORY (explicit in the OBJECTS clause) because
there are NO underlying attributes currently defined in the PWG
Job Monitoring MIB v1.0 ('xxx-completed' attributes are defined,
but the 'xxx-requested' ones are not, except for the one
'jobKOctetsTransferred' which should have the same value as
IPP's 'job-k-octets').
DC> What about
DC> jmJobKOctetsPerCopyRequested,
DC> jmJobImpressionsPerCopyRequested, and the
DC> sheetsRequested attribute?
DC> Aren't these the "underlying attributes" you're looking for?
DC> I think the first of these two are even MANDATORY.
I'm trying to propose new notification groups and object groups
for addition to the PWG Job Mon MIB.
Are you suggesting that these missing job attributes be defined
in a revised PWG Job Mon MIB?
DC> I think they're already there.
The real problem with gas gauges is that while job size in
k-octets of raw PDL may be known, job size in impressions
is almost never known before rendering. Further most PDL
interpreters I've seen don't even have hooks for reporting
progress out to the job scheduler/SNMP agent interface. You
can't instrument what doesn't exist. Maybe IBM writes their
own PDL interpreters? Most printer vendors buy them from
third-parties.
DC> I would think that most printers, whether they have hooks
DC> into the PDL or not, would have the ability, at least if they
DC> wanted to, to keep track of how many impressions the PDL
DC> is spitting out. That is, the printer has to print the
DC> impressions, so it is not unreasonable to imagine that it
DC> could count them as well.
DC>
DC> As far as the not "known before rendering" issue, we've
DC> discussed this before, and my contention is that knowing it
DC> at some point is better then never knowing it. If you, Ira,
DC> look at a queue and see that John Doe is printing a job whose
DC> status is "Page 12 of 27", that gives you a much better idea
DC> of whether you want to print to that printer than if the
DC> status was "Page 12", even if only a few moments before, the
DC> status was simply "Page 10" due to the total impression
DC> count 27 not being known yet.
Cheers,
- Ira McDonald, consulting architect at Sharp and Xerox
High North Inc
-----Original Message-----
From: dcarney at us.ibm.com [mailto:dcarney at us.ibm.com]
Sent: Friday, March 31, 2000 4:05 PM
To: McDonald, Ira
Cc: harryl at us.ibm.com
Subject: Re: IPP> FW: Slides for Job Mon MIB trap extensions for PWG-C
Ira,
I'm sorry to be coming in late in the game here, but I have looked at your
proposal at a high-level and I'm wondering if the "gas gauge" idea has been
forgotten.
Specifically, in
ftp://ftp.pwg.org/pub/pwg/ipp/new_NOT/draft-ietf-ipp-not-spec-02.txt
page 42, attributes 'job-k-octets', 'job-impressions', and
'job-media-sheets' are optional on 'job-progress' as well as
'job-completed' notifications. I remember we (you and I on the mailing
list) had a spirited discussion about this, but in the end, the attributes
got added to the notifications.
However, in
ftp://ftp.pwg.org/pub/pwg/ipp/new_NOT/draft-ietf-ipp-not-over-snmp-02.pdf
page 14, those same "attributes" are missing from the jmJobProgressV2Event.
Is there something I'm missing (it wouldn't be surprising since I have been
absent from all the discussions of this!)?
In any case, I would think that if the PWG IPP group decided to put these
attributes in IPP notifications, they would also decide to put the same
attributes in SNMP traps meant to convey IPP notifications. No?
Dennis