Leslie,
Thanks for replying - .
I tried changing the Config and Status polling intervals for the collection, but
the objects still remained. I received another suggestion from Jeff Fitzwater to
try using the xnmsnmpconf command to clear the cache and to issue it again to
force a new resolution. Once the xnmsnmpconf cache was reset, the new polling
intervals took effect and the nodes were delete.
As for my "real" problem -
As my note indicated, our remote offices are constantly moving and it is
difficult keeping up with the changes. While a better change process would help,
the situation originates in the business unit. For instance the business unit
may decide to build a new office, but decide to staff it by consolidating other
nearby offices. Unfortunately, we won't become aware of this decision until
after the staff has already been relocated and the facilities removed.
Since Netmon automatically discovers the new location, we can begin managing the
new facility immediately. (Even before the office is "officially" placed in the
appropriate submaps.) However, old offices tend to remain on the map for several
days. Since Operations is usually the first to learn that an office has been
abolished, I've written a script that they can use to send email messages to the
appropriate parties confirming that the office has been closed, validate that
the interface cards within the hubsites have been shutdown, and then build a
collection of all network nodes that need to be eliminated.
Because the Operations staff isn't trained to perform map management, I don't
want to give them access to the read/write maps. Additionally, our map
administrator complains that it can be very time consuming locating the office
and link icons that need to be deleted. Thus, he usually updates the map only
once or twice a week. My first thought was to reduce the work effort by having
a collection of everything he needs to delete. Then he could go to that
collection, scan the page for errors and perform a "Select All" and "Delete from
ALL" very easily. As an after thought, I though about eliminating the task
all together by changing the Node Down Delete Interval for that collection.
Here is how it works. When Operations learns that an offices has been abolished,
they type: "removeoffice -r xxx" where xxx is the 3-char prefix of the primary
router in that office. From there, the script builds the collection based on the
networks attached to that router and confirms that everything is down. Once it
is part of the collection, it can be eliminated from the network maps using the
Node Down Delete Interval. Since it is the script builds the collection and
knowing that NetView isn't going to accidentally delete anything that is
actually UP, I feel comfortable giving the script to a larger user population
than just the map administrator. So far, the script has been very easy to write
and hopefully it will allow us to keep the map more up to date.
Brad Martin
MetLife
"Leslie Clark" <lclark@US.IBM.COM> on 11/13/99 05:45:53 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: Brad Martin/Bsg/MetLife/US)
Subject: Re: node down delete interval (was blank)
Brad, the fact that the node was deleted overnight confirms James'
proposition, that the node down delete interval is related to the
configuration poll interval. A demandpoll and a configuration poll
are not the same thing. Your config poll interval is probably the
default, which is one day. If you examine the output of demandpoll
you will see that it includes the timestamp for the next configuration
poll for that node.
The design supposed that you do not want to do config polls too
frequently since they generate a fair amount of network traffic. The
expectation is that the node down delete interval will be longer than
the config poll interval - hence the defaults of 1 day and 1 week.
In real life, though, I find that customers generally delete things
manually
when they know they have been removed from the network. Otherwise
you are might well be masking a real network problem. So I don't
think automatic deletion is used too much. Anybody else?
So now, just to be nosey, let me ask what problem you are trying to
solve with this rapid deletion of down nodes? Maybe someone will
have a suggestion for an alternate solution.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
James,
Sorry about forgetting to include the subject and releases. NV 5.1.1 on an
AIX
4.2.1
It appears that the nodes will eventually get deleted but I don't
understand
what triggers it.
I've done a demand poll and the objects remain.
I've closed and re-opened the user interfaces - the objects remain.
I've done an ovstop / ovstart - the objects remain.
I've gone home and returned - the objects disappeared.
My goal is to find an easy way for Operations to remove remote offices that
have
been relocated. (Averages 20-30 offices / month. The new office is
typically
created using new addresses / names and then the old office equipment
recycled).
I've written a script that determines the objects that belong to an office
based on the router in the remote office. The script updates a collection
called
"Garbage" with the appropriate rules. I can have the map administrator
manually
remove these items, but it would be ideal for NetView to then delete these
objects based on the Node Down Interval. Using this technique, I can give
the
script to operations knowing the worst they can do is run it for an active
office and even then NetView isn't going to delete it unless the office is
actually down.
Brad Martin
MetLife
"James Shanks" <James_Shanks@TIVOLI.COM> on 11/10/99 05:17:06 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: Brad Martin/Bsg/MetLife/US)
Subject: Re: node down delete interval (was blank)
I'm not the netmon guy so I am just guessing. But my understanding is that
the
node down delete thing would only be checked when a config check was done.
Have
you tried demand polling the thing?
James Shanks
Tivoli (NetView for UNIX) L3 Support
Brad Martin <bmartin@METLIFE.COM> on 11/10/99 03:51:35 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:
If I understand the Node Down Delete Interval in the SNMP Configuration
Dialog
correctly, I should be able to create a collection that has a node down
delete
interval that is very short. For instance, I should be able to place a
router
that has been critical into a collection and then specify that nodes within
that
collection are to deleted if they've been down for more than 30 minutes.
However when I create a collection with one node that's critical and then
configure the SNMP node down interval to 5 minutes, nothing happens. Does
anyone
have an explanation?
Thanks,
Brad Martin
MetLife
|