On Unix, there is
an object created for the service on a node, and that object has a status
field. The object has a name like NodeName:ServiceName. A trap is
generated when that status changes. Graphical representation of the presence
of the service is provided by means of Smartset membership based on the field
set on the node at discovery of the service. The only thing missing is
graphical representation of the status of the service object.
This is exactly the same as it was
for nvsniffer in v601. The work that has gone into this has focused,
apparently, on making the monitoring more robust and more sensitive to
smartset membership so that it can play a real role in out-of-the-box
monitoring - for the event-driven world.
If you do 'ovobjprint -s | grep NodeName', you will see among the usual
objects for the node, the object for the service. Now who can come up with
some tricky little thing for representing their status graphically? I'm
picturing an APM-like topololgy.
Cordially,
Leslie A. Clark
IBM Global Services - Systems
Mgmt & Networking
Detroit
|
| "Barr, Scott"
<Scott_Barr@csgsystems.com>
Sent by: owner-nv-l@lists.us.ibm.com
12/05/2003 10:37 AM
Please respond to nv-l
|
To:
<nv-l@lists.us.ibm.com>
cc:
Subject: RE: [nv-l] servmon and
itmquery in NV 7.1.4
|
Jane, could you tell us how you REALLY feel?
(ducking)
My PMR is essentially dead. I will share the response I got
from Development and I will also share my response to development. I'm sorry,
folks are going to dislike what I am about to post. Unfortunately, you all
won't know anything about servmon or it's status if I don't since apparently
it's not going to get any attention from anyone anytime soon.
I just want to re-iterate to everyone
listening - level 2 support team ROCKS! (And I mean all of you).
First Response to the PMR
-----Original Message-----
From:
(removed)
Sent: Wednesday, November 05, 2003 1:30 PM
To: Barr,
Scott
Subject: RE: PMR 13988
Scott,
I got confirmation that
servmon will not draw icons for the services it
discovers, it will only
populate a smartset with devices that have the
service discovered.
The
multi-homed boxes should get drawn in the smartset. I will test
your
mutli-homed scenario by turning off the service from the main
snmp
interface and then seeing if it populates the
smartset.
(removed)
My Response
To: (removed)
cc:
Subject: RE: PMR
13988
Do I understand correctly that Windows
versions DO draw the icons?
Here is my problem: My mainframes have
20-30 services running on them. My
main application servers have as many as
100 services running on them. If a
service fails, I have no way of know
which one failed. The icon shows up in
the smartset saying "I Have those
services you want to know about" but I
have no way of knowing which
one.
This seems like a pretty useless feature
without that information and we've
been patiently waiting for it since
nvsniffer made it's debut. Please tell
me service icons can be created to
reflect which services are down. And if
you tell me it works on Windows,
and not on Unix, I'm going to get down
right obnoxious since that screams
"the code wasn't done yet".
Sorry to be a pain, but we really really
needed this feature and I am quite
dismayed if it isn't actually
useful.
Another question, can you provide
details about what kind of code is
required for the "discovery test" i.e.
does it have to be java? Is there an
archtected return code or something?
More details would be very helpful.
Next Response from Support
From: (removed)
Sent: Friday, November 07, 2003 3:01
PM
To: Barr, Scott
Subject: RE: PMR 13988
Scott,
Windows version does draw
icons for the services and will display status of
the service. However, it
is based on node and not interface, so 3
interfaces each with a service
will only display 1 icon.
I don't believe there is a way to create icons on
unix so as to reflect
which services are down. I will need to double check
this with L3.
I understand your position and frustration.
I will also
check on the discovery test issue you raised.
(removed)
To: (removed)
cc:
Subject: RE: PMR
13988
(removed) any update? We have a comittment from an application team to
write a
status program.
Does it seem to you that the code was
maybe released before it was done? I
really like/need the functionality of
servmon but I am sorely disappointed
it is so crippled in this release. I'm
not a big fan of "wait until next
release" so with there actually being
people wanting to use it, what do you
think my chances are of getting one
that is functional?
The status application specs would be
appreciated.
And here is the "*FINAL*" response from
support
Scott,
I have checked with L3 and
there is no way to create icons for the
services.
WARNING THIS WAS MY FINAL RESPONSE TO
SUPPORT - FLAME ON
First off, let me assure you that I in
no way feel that your support has been anything but perfect. NO L2 complaints
here.
Okay. Now, ask them this...
Why not?
If that sounds fascetious I'm sorry. I
think this is a bunch of crap.
Second off, I am once again severely
disappointed in L3. I think once again, a feature was included in a release
that was not ready to be delivered. For goodness sake, if I put my mind to it,
I guarantee that I could write rulesets and database automation to build the
services in the map. It would be a time-consuming, resource-wasting
administrative-nightmare kludge or I would have done it already. But it could
be done. The answer "there is no way to create icons for the services." is
obviously incorrect. Of course they could. They don't want to, have "more
important" things to do (Maybe they are working on my 2 year old enhancement
request to allow smarset icons in web client scopes, but I doubt it). They
have icons for everything else. The windows version has icons. Nothing in the
manual/release notes indicates I should not expect icons for the services.
Why do I need icons? Well, to be honest,
our mainframes and application servers have dozens of ports in use. Not so
many that icons would be prohibitive, but enough that navigating a menu drop
down doesn't help much and a visual cue would be much better. (Especially
since the entire application is driven around a visual entity - the
smartset).
I am sure there is zero chance of this
PMR being closed with me being satisfied so you can close it as "L3 won't fix
it and the customer doesn't have any say" Feel free to send it up the
escalation chain. And people wonder why IBM occasionally gets a black
eye.
I'm not bitter, just angry and vocal.
This is the "new netview" where the emphasis is on deploying new features as
fast as possible and leave old features broken or castrated (NetView security,
native console, "backup" functionality etc. etc.) I would be happy to discuss
this new "feature" with anyone internally who is interested. And, once again,
I volunteer my name for NetView pre-release/beta/early support testing. In my
estimation this code is broke.
- Scott
Aren't you glad you asked.