See below. I of course think you ought to just upgrade and try it.
You can always turn it off.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit
"Boyles, Gary P" <gary.p.boyles@intel.com>@tkg.com on 09/20/2000 10:30:51
AM
Please respond to IBM NetView Discussion <nv-l@tkg.com>
Sent by: owner-nv-l@tkg.com
To: "'IBM NetView Discussion'" <nv-l@tkg.com>
cc: "Boyles, Gary P" <gary.p.boyles@intel.com>
Subject: RE: [NV-L] Unreachable Status.
Leslie,
Thanks for the help. Three more questions (I'm still
confused about a couple of items)...
1) If the router-interface goes down, and a server
on that subnet is polled (normal polling cycle),
will I receive an "Interface Down" event?
> Not unless you have some other way to get to that
> subnet. Like another router whose interface on that
> subnet is up. The normal polling cycle, as you call
> it, does not apply. You should not be able to reach
> the server, whether it is up or down, if no router
> has an interface up on that network. However, there
> is a known phenomenon when you have not discovered
> and mapped all of the routers on that subnet, and
> it is marked unreachable, but you still have some
> unknown (to netview) path to it. This unexpected
> ability to reach the subnet causes rapid status
> changes - like a blinking xmas tree. Normally the
> philosophy is that when the subnet is occluded, we
> must make no assumptions as to the state of the
> other nodes on that subnet.
2) If the subnet is still unreachable, and I force
a poll (e.g. demand poll), will this generate
any additional event?
> Don't know.
3) If a node (server) has two NICs, and is on two
subnets... but one subnet goes unreachable -- will
I get a "node unreachable", "node marginal" +
"interface down" ... or something else?
> Don't know.
Again, thanks. Your explanations have been a big help.
They should be rolled up and put in the NV documentation
(really... they should).
Regards,
Gary Boyles, Intel
-----Original Message-----
From: lclark@US.IBM.COM [mailto:lclark@US.IBM.COM]
Sent: Tuesday, September 19, 2000 7:26 PM
To: IBM NetView Discussion
Subject: Re: [NV-L] Unreachable Status.
Comments below.
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
Detroit
"Boyles, Gary P" <gary.p.boyles@intel.com>@tkg.com on 09/19/2000 07:24:02
PM
Please respond to IBM NetView Discussion <nv-l@tkg.com>
Sent by: owner-nv-l@tkg.com
To: "'IBM NetView Discussion'" <nv-l@tkg.com>
cc: "Mauch, Mike" <mike.mauch@intel.com>
Subject: [NV-L] Unreachable Status.
To: NetView Forum.
I have a few questions on "Unreachable" status as defined
in NV (NT) V6 docs.
1) Is this working, or does it have to be purchased?
I had heard that it was now free, but is that in V6 or
V6.01?
> It is available in V6.0, to be turned on with a script
> you can get from support. It is turned on by default in
> 6.01. It can be turned off.
2) Assuming its a part of NV... does it mark the
interface as unreachable if the next-hop (router iface)
is down, or does the node have to be totally unreachable
(i.e. all interfaces off all subnets are unreachable).
> Subnets are unreachable if no managed router interfaces on
> them are reachable. Nodes on that subnet are unreachable
> by decree. Routers are unreachable if all subnets they
> touch are unreachable. If we can get to any interface
> of a router, it is some other status (marginal, etc).
3) Is the status marked "Unreachable", or is there an
additional "unreachable" NetView trap.
> The IP Status is in fact set to Unreachable and the color
> is changed to white for routers,networks, and router
> interfaces only. For performance reasons, the color of
> the rest of the nodes on an unreachable subnet is left as-is.
> Their unreachable status is indicated by the white router on
> that segment submap, or by the white color of the network
> icon. There are new traps as well. The 'Router Down'
> or 'Router Marginal' trap is the root-cause event. Open a
> ticket, send a page when one of these comes in. The
> 'Network Unreachable' trap is your indication of the scope
> of the outage. There are no further events about the down-stream
> devices until the subnet is reachable via a router interface again.
4) If I'm managing 100 server off of a subnet, and the
subnet goes off-net... will I still receive 100 server
I_DWN events, or are they suppressed? If they are suppressed,
how do I indicate an "off net" event for a server?
> The product indicates the EXTENT of the outage only
> graphically, as far as I can tell, unless you know which subnets
> encompass which nodes. You see a part of the map go white, and
> that's what is affected. You get a Network Unreachable event,
> and it is up to you to know that the servers are on that subnet.
> Further status events about those servers are suppressed until
> netmon can reach some router, any router, on that subnet.
>
> The goal here was to provide suppression of downstream events
> WITHOUT having to maintain a knowledge of all possible routes
> within the product, making it very lean. The approach taken
> was to rely on the OS and the network, which may be rerouting
> dynamically. You have to try it. It is very cool. But root-cause
> analysis may be an inappropriate description of what it does.
----------
Basically, I'd like to deliver the best possible (root-cause)
message for the available data.
Regards,
Gary Boyles, Intel
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
_________________________________________________________________________
NV-L List information and Archives: http://www.tkg.com/nv-l
|