What you describe is not exactly a test of the problem, although it does
point out another example of the limitations of r/o maps. Manage/unmanage
is not the kind of status that can be reflected on a r/o map, since you
need a r/w map to do it. It is expected that a refresh must be done to see
this on a r/o map.
A valid test is as James described, a managed device whose actual status
changes between down and up, and there is only a r/o map up at the time
that the change takes place. If the product is working correctly, then the
change will show up at the r/o map. The question is whether this is somehow
inhibited by certain methods of remote access, such as x-station or
x emulation, or networkstations. There seems to be some anecdotal
evidence that it might be, however many of the cases cited seem to involve
situations that would not work anyway because they involve functions that
are not expected to work on a r/o map regardless, such as changes in
manage/unmanage, or add/delete/move of symbols.
The design of the product is such that IP status changes of permanenet
symbols do flow to r/o maps without the presence of a r/w map.
When you decide on your architectural design, you need to understand
the limitations inherent in r/o maps and weigh them against the
administrative
burden of multiple maps. It depends on the manpower you have for
administration and the needs of the operators, among other things. Every
implementation will be a little bit different.
The implications of r/o maps are not well documented, but are easy to
deduce, since they all flow from the same basic truth: You cannot add or
change or move or delete symbols on a r/o map, and neither can the
product daemons or processes. You can ack/unack, but these are local
only to that copy of the map. Everything else must be done on the r/w map,
and the r/o must be refreshed to see it.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit
==============================================================
I would like to throw in my 2 cents if I may. I still consider myself a
novice
at supporting
Netview but this is what I did to check this out.
I had my r/w map open on my X-station at my desk. I opened an r/o map on
the
X-station in our
command center. On the r/w map I managed an interface on and AIX server I
knew
would turn red.
I checked the r/o map every couple of minutes for about 15 minutes. No
status
change appeared
until I did a refresh on the map. Then the interface I had managed showed
up
red. No HSRP involved
here.
Next I unmanaged the same interface. Brought up an IP map through the web
interface. Re-managed
the same interface. Within a second of the r/w map showing red the same
interface on the web
session showed up red.
Until this came up on this forum I thought it was just the way the r/o map
worked. So I created
another r/w map for the Command Center here, so their map would dynamically
update.
James, are you saying that I should open an PMR on this situation ?
Michael Jaedike - IT&S
Venator Group
I just posted a re-create scenario. Does it work for you or not?
James Shanks
Tivoli (NetView for UNIX) L3 Support
Dean Grant <dgrant@AHOLDUSA.COM> on 02/03/2000 02:48:48 PM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView
<NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: James Shanks/Tivoli Systems)
Subject: Re: NetView over XWindows?
I don't know if its related to our environment or what, but we noticed that
when
we upgraded from NV AIX 5.1.1 to 5.1.2, status updates did not occur in a
reliable fashion on the R/O maps if there was no R/W map open. Network
Operations will notice a difference (and call) in about 20 minutes if the
map is
not open. There must be some correlation there. It definately started with
5.1.2
James Shanks <James_Shanks@TIVOLI.COM> on 02/03/2000 12:43:37 PM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView
<NV-L@UCSBVM.ucsb.edu>
To: NV-L@UCSBVM.ucsb.edu
cc: (bcc: Dean Grant/AIS/US/Ahold)
Subject: Re: NetView over XWindows?
There are two issues here, not one.
(1) How does NetView work over a normal X GUI, and
(2) How does NetView work over a particular emulated X GUI.
I can answer the first. I don't know about the second. We do not use
emulators
on a regular basis, though some individuals do, and we do not test the
product
with them. Our job is done when the X GUI works. It is up the particular
brand
of X emulator to reflect that on the emulated display.
Do you need a read-write map open to see status updates on a read-only map?
The answer is "no".
For an already-discovered object, present on the map at the time it opened,
the
icons will change color as status is updated. I just tried it with the
5.1.2
boxes on my desk. I opened the only map read-only (netview -ro), did a
locate
and opened the segment submap and the collections submap (the collection
was on
SNMP-supported objects), and saw my other box (this one I am using to send
notes
with). It was green and normal. I simply shut it down and watched the
other
box, the one with the open read-only map, and waited for the events to come
into
the event display which indicated it was down. The icon on both submaps I
had
opened turned red as expected when the events were in the display. I then
powered up the down box and shortly thereafter netmon noticed it, and once
again
I got the Up events and the icons turned green at about the same time.
This
proves conclusively that status will propagate correctly to a read-only map
when
there is no read-write one opened.
Now, if this isn't what you see on your X GUI then you have a problem that
you
should notify NetView Support about. Note that I am not talking about
devices
discovered after your read-only map is opened, or about devices coming and
going
into collection submaps based on status. A read-only map will not
re-position
an icon without a refresh, but if we are talking about just the color of an
existing object, this should propagate fine.
Hope this helps
James Shanks
Tivoli (NetView for UNIX) L3 Support
Leslie Clark <lclark@US.IBM.COM> on 02/03/2000 10:58:50 AM
Please respond to Discussion of IBM NetView and POLYCENTER Manager on
NetView
<NV-L@UCSBVM.UCSB.EDU>
To: NV-L@UCSBVM.UCSB.EDU
cc: (bcc: James Shanks/Tivoli Systems)
Subject: Re: NetView over XWindows?
About this business of needing a r/w map open for r/o maps to work:
I have heard this rumor here before, but I have trouble believing that it
is working as designed. Has anyone taken it up with support and been
officially notified that a stand-alone R/O map will not reflect status
correctly? I am just about positive that this has not always been the case,
and I don't see it documented as a feature....
If I were not getting status updates, I would suspect that the ipmap
process
for that user had failed, possibly due to insufficient ulimits.
I would appreciate word from anyone who has been given this information
by Support, as opposed to just observing what works and living with it.
(Now watch me get in trouble with Support for asking questions like
this....)
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit
===========================================================
Mark,
If you do not have read/write access to the map it will not update unless
you do a map refresh.
-----Original Message-----
From: Discussion of IBM NetView and POLYCENTER Manager on NetView
[mailto:NV-L@UCSBVM.ucsb.edu]On Behalf Of Mark Marcelissen
Sent: Wednesday, February 02, 2000 3:06 PM
To: NV-L@UCSBVM.ucsb.edu
Subject: NetView over XWindows?
I am attempting to have a NT client off a NetView for UNIX using Exceed.
I am sending the DISPLAY back to my NT desktop from the NetView for UNIX
server
, it displays OK and the Menus work, BUT, the maps or submaps do not update
with status changes. For example: Nodes is a critical state.
A FILE-REFRESH MAP updates the screen.
A NetView session is already running when netview is executed from the
command line.
Has anyone attempted this or is this a known problem?
Mark M.
DFAIT
|