Re: [nv-l] More location.conf questions...

[Leslie Clark <lclark@us.ibm.com>]

Chris, I'm not really clear on what you are saying you are doing, and I'd like to. Let's make sure we are talking about the same thing first. 

Here's a concept to understand: (this may already be well understood, but that's never stopped me before:)  A location submap has the same properties as the IP Internet submap. That is, it can only contain routers, networks, and other location icons. These are the same things that you can specify in the location.conf. A router appears at this level, and so do subnets for all of its interfaces. The router symbols you see WITHIN network submaps are placed there by ipmap automatically, not because of anything you specified in location.conf, the same way hubs and servers and printers appear in the segment submaps.

So you have a new router. To get it completely put away, you need to put these things in location.conf:

1) For each addressed interface, you need a NETWORK entry, which has 3 or 4 fields:
That is:  LocationName NetworkAddress LocationType <ParentLocation>
Each of those networks will contain a router symbol representing its interface on that subnet, automatically. But what you specify in location.conf is the network address (eg 192.168.20.0) of the interface, NOT the interface address. 

2) The top-level router symbol will automatically be placed in the logical place between the location submaps. Depending on where they are, that may mean that the router is placed on the IP Internet submap. If all of the interface subnets are in the same location submap, then the router goes in there as well. If you want it to be placed somewhere else, you may add a GATEWAY entry, which has only 2 fields:
That is: LocationName RouterSelectionName 
The doc actually says that you don't have to use the Router Selection Name, and that any address will do. I have not found that to work. In any case, the expectation is that you would only make ONE such entry for a given router. If you make more, you can expect unpredicable results. I would expect it to use the first one or maybe the last one. Even if you don't place all networks, you can still place the router.

I would consider using the rnetstat -I command against a router address to get a list of the networks that will have to be placed, independant of your DNS. And I would be careful to use only 1 gateway entry. If you use loopbacks, you should be able to do this neatly.

I hope this helps. 


Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
(248) 552-4968 Voicemail, Fax, Pager

"Christopher J Petrina" <cjp8@meadwestvaco.com> Sent by: owner-nv-l@lists.us.ibm.com 02/25/2004 09:46 AM Please respond to nv-l

To:        nv-l@lists.us.ibm.com         cc:                 Subject:        Re: [nv-l] More location.conf questions...

Doing it the old fashion way is a way to do it.  For myself and some others, I think what we are looking for is a more automatic way.   We do not have people that are constantly watching the netview box, when our support teams needs to get to a device they will open the map and find it.   
    The way we are populating Netview is the following: 
       We have a DNS database that is dumped nightly. We look for new devices in the nightly dump and compare them to the existing dump from the night before.  Anything new is placed in a loadhost script, anything not in the new DNS but was in the current (old) one is then placed into a delete script (nvmaputil.sh) <= we are not waiting for the node down to delete interval to pass.  We have city and states in the database that we strip out, and based off that is where the device is to be placed in the location containers.   
    Our problem occurs when, if we do not have all the interfaces properly listed out then the location.conf will fail and place the device in both the location container and on the IP Internet map.  We are only trully concerned with Routers, Switches, Servers and some RF devices.  Since the locations are primarily about networks,  I strip out all the subnets and use those to create the city/state containers and then use the routers as Gateways for each container.  Our DNS contains all the devices and their interfaces.  So the same device (lets say a router) actually has many multiple entries.   How does Netview deal with finding a router many different way after it has already discovered it?   Does netview find a device and then if it finds an interface on that device again, does it rename it or simply say that the device was already found another another name?  I think I have seen the warn! ings before in location.log about!
  that. 

Thanks 
Chris Petrina

Leslie Clark <lclark@us.ibm.com> Sent by: owner-nv-l@lists.us.ibm.com 02/25/2004 12:39 AM Please respond to nv-l

To:        nv-l@lists.us.ibm.com        cc:                Subject:        Re: [nv-l] More location.conf questions...

You are being caught by that next section of the header file where it says 
#              Overlapping Ranges are unpredictable.  The following 
#              is invalid: 
#                Tivoli       146.80-90 State 
#                SalesOffices 146.75-95 City Tivoli 

You are using implied wildcards on your NETx entries, but your catch-all is an overlapping range. I know you could read this doc to mean it should do what you want it to do, but it does not actually work that way. I had a customer take it up with support a while back and they clarified that it works the way it works. You will have to be a little more specific in your catch-all definition, I'm afraid. 

Actually, I get the impression that as a group, new users have the idea that the only way to configure the map is with the location.conf file. There is still the old-fashioned way - cut and paste. For what it's worth, here's how I use location.conf when I'm implementing Netview for a new customer network. 

1) First discover the network at least once, and at least all of the routers. Do this to make sure they are all connected properly, and named they way you like them. Look at them in a Smartset if they make a furry black ball:) Putting them away too soon can mask configuration errors that Netview would otherwise show you. 
2) Make a location.conf file that builds the whole location hierarchy for you,  and places the routers, and the easiest parts of the network ranges. Concentrate on the remote stuff. That will thin out the map the most. Think about leaving the core stuff right out on the top layer. 
3) Fiddle with this a bit, using file....new map (get efix for apar IY48698) until the map is at least readable with zoom 
4)Rediscover using the location.conf 
5)Take the location.conf out of effect (rename it), then close/open the map 
6)Finish it up by hand. You will change your mind a lot about the final details, and it is not worth fighting against the location.conf every time you do. Add some temporary locations manually to hold any large fans that you are not ready to deal with just yet. 
7) When you think you are done, update the location.conf to match, in case you need to do a rediscovery later on. You can do this manually, or ask around. A number of people have come up with scripts that will generate a location.conf from an existing map. 

A design consideration: A map with three things on it does not convey very much information except when everything is up. I like to see the core infrastructure right on top - eg all of your pairs of core switches and the major subnets between them. Then when something does turn yellow, you can gauge the impact at a glance. 

Cordially,

Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit

"Duppong, Jason" <jason.duppong@thomson.com> Sent by: owner-nv-l@lists.us.ibm.com 02/24/2004 03:33 PM Please respond to nv-l

To:        "'nv-l@lists.us.ibm.com'" <nv-l@lists.us.ibm.com>       cc:               Subject:        [nv-l] More location.conf questions...

Hello List, 
I don't want to wear out my welcome to this list and I really hope I'm not the only Netview newbie that gets tripped up on the location.conf file, but I'm afraid I still need a bit more clarification.  I really want to understand exactly how Netview uses this file for placement of objects.  To this end I've read and re-read release notes and most of the Unix documentation for Netview, searched through archives on this mailing list, and played with the file in general just trying to figure things out.  When I make a couple changes and expect one thing and those changes are not reflected on the MAP the way I understand I get really confused.  I have a rather complex location.conf file, about 600+ lines, breaking our companies various subnets into containers and sub containers based on how our network is logically setup (I've worked with countless network personnel in coming up with the rules for this file).  Where I'm conf! used right now is how to mak!
 ! e an "INTERNET" container (doing this is better then hiding 300+ nodes and segments from the IPMAP).  According to the examples for location.conf, point D indicates: 
 
#           d) If a network matches more than one entry, the most specific
#              match will be used.  So, if we were matching the network
#              146.84.5.5, the address patterns that it would match (in order
#              of most specific to least specific) are the following:
#                146.84.5.5
#                146.84.5
#                146.84.1-6
#                146.84 
They way that I understand this example is that I should be able to do an "INTERNET" container with a line similar to the following: 
 
INTERNET    1-254    U.S.A 
 
When I restart map generation the initial symbols on my MAP look real promising.  Even during the regen the number of symbols on the map remain relatively constant except for a number of lines linking to INTERNET that shouldn't.  Then when the map finishes drawing and I double click on the INTERNET container, I find, to my surprise, all the nodes that should be in a different container (in this particular case it is all the internal corporate IP's to my company, which explains all the lines from the other containers).  I have placed the INTERNET container at the top and bottom of location.conf, nothing makes a difference.  Presuming my companies internal subnets were 10.1.x.x and 192.168.x.x, wouldn't a location.conf looking like this work? 
 
NET1    10.1        Site2 
NET2    192.168   Site3 
 
INTERNET    1-254    U.S.A 
 
Based on the example, Netview would break out all nodes in the 10.1.x.x subnet and the 192.168.x.x subnet and place them in the NET1 and NET2 containers while placing all other segments and connections between NET1 and NET2 to INTERNET if those links/connections exist?  I understand that there will be some stragglers left over that will appear on the IP Map, but is what I'm trying to do a valid use of the rules for location.conf?  I understand that I probably have a few typos in my location.conf file, it is 600+ lines long, but this is what the location.conf file is for right, describing your network?  My network falls into the Medium category for size, so hopefully someone else has a huge location.conf and already has this working.  I guess what I'm really after is a confirmation that this should work, that way at least I know I'm not in some wild goose chase. 
 
I don't think I can thank you Netview Gurus enough for the comments posted since I've joined this list.  I hope you understand how valuable your experience and comments are to us Netview Newbie's :) 
 
Thanks in advance for any responses.... 
  
Jason