I've seen similar problems with multiple interfaced devices. And, as
you say, a demand poll doesn't always correct status color, even when
all of the interfaces are up. In fact, in some cases it doesn't even
help to manage/unmanage, which should trigger a poll. Some cases, only
way is to go to the interface map and unmanage/manage individual
interfaces. Is it possible that you have secondary addresses or
aliases, and don't have that netmon flag turned on? Perhaps the poll of
the interface table doesn't work for aliases if the secondary address
flag is off.
Are you running any third party applications that may propogate status
based on other criteria (routing protocols, bridging, CPU utilization,
or some such)than just interface status?
Just a thought.
Jim Kellock
---------------------------
Patrick McNeil wrote:
>
> They are Cisco devices with SNMP enabled and working properly. I only
> noticed the problem on a series of routers that are deployed
> internationally. It is a Cisco 3640. We have several other ones deployed
> around the US with no problems. There are two with the same configurations
> deployed internationally (several more with direct connections back to the
> US).
>
> One of them displays this behavior more than the other.
>
> I know it is difficult to troubleshoot over email with an intermittent
> problem, I was mainly hoping someone else has come across this problem and
> could say tweak this parameter and the problem would go away.
>
> Patrick
>
> -----Original Message-----
> From: Discussion of IBM NetView and POLYCENTER Manager on NetView
> [mailto:NV-L@UCSBVM.ucsb.edu]On Behalf Of Dean Grant
> Sent: Tuesday, February 22, 2000 2:11 PM
> To: NV-L@UCSBVM.ucsb.edu
> Subject: Re: Weird map update problem
>
> What kind of routers/hubs are you using? Are they enabled for SNMP?
>
> We had a very weird polling/map update problem a couple of years ago where
> NetView was reporting false downs for routers. Working with the support
> center
> and using sniffer traces we were able to determine that for non-SNMP devices
> that are known to be up, during the polling cycle NetView issues an ICMP C
> Get
> Address followed by an ICMP Echo. If a non-SNMP node is known to be down, it
> issues only an ICMP Echo. The routers in question, which were Hypercom, did
> not
> support the ICMP C Get Address request. This is a fault in their IP stack
> implementation, not NetView.
>
> You problem does not sound exactly like what we saw, but very similar. If I
> were
> you I would call support,. It was very difficult for us to troubleshoot and
> with
> out the NetView support desk I doubt we would have found it.
>
> FYI, we fixed the problem by adding a un-documented flag to the
> /usr/OV/lrf/netmon.lrf file.
>
> Hope this helps
>
> Dean Grant
> Ahold Information Services
>
> Patrick McNeil <patrick_mcneil@YAHOO.COM> on 02/22/2000 02:29:29 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: Weird map update problem
>
> To answer the questions:
>
> 1) When I demand poll the device, the status does not change (I demand
> polled on the network layer, the one below that and the segment layer). All
> of the interfaces on the device are green when you drill into the router
> (there are 9 of them if that matters).
>
> 2) When I unmanage the router, it turns wheat. When I remanage the router
> it turns back to green.
>
> 3) It is actually all of the interfaces on the router are green. Each of
> the submaps that contain the router are yellow, with the router being red in
> the submaps. The trapd.log file does contain the interface down message and
> the interface up message for the interface that went down.
>
> Unfortunately, I cannot recreate the problem on demand (which I know makes
> it difficult to troubleshoot). It happens about once or twice a week where
> it says red on the submaps, but not on the network map. This network goes
> down typically one to two times a day (carrier problems not our problem) and
> most of the time it is fine. Just recently (since about the end of
> December) has is started doing this. The only thing we did around that
> timeframe was stop the system for the Y2K backup and restart (no patches
> applied or services added).
>
> Patrick
>
> -----Original Message-----
> From: Discussion of IBM NetView and POLYCENTER Manager on NetView
> [mailto:NV-L@UCSBVM.ucsb.edu]On Behalf Of James Shanks
> Sent: Tuesday, February 22, 2000 11:53 AM
> To: NV-L@UCSBVM.ucsb.edu
> Subject: Re: Weird map update problem
>
> You got me. But there are three questions I would ask to try to determine
> what
> is going on.
> (1) If you go to a submap where the router is red, and demand poll it, does
> the
> color change?
> (2) If you unmanage it, it will turn wheat colored. When you then manage it
> again, does it turn green again or go back to red?
> (3) On a submap where the router is red, determine what inteface that is.
> Can
> you find a matching Interface Up trap in trapd.log?
>
> James Shanks
> Tivoli (NetView for UNIX) L3 Support
>
> Patrick McNeil <patrick_mcneil@YAHOO.COM> on 02/22/2000 09:43:45 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: Weird map update problem
>
> We have been having some problems with maps updating on our NetView system.
> Our system has the following specifications:
> 4 processor F50
> 3 GB Ram
> 1 GB Paging Space
> Lots of free disk space
> Netview for AIX 5.1.1
>
> We have a central hub site internationally that has six locations that
> connect to the hub for corporate connectivity. When the hub site goes down,
> all the sites behind it goes down. We get the appropriate messages from
> Netview that everything is down. When the site comes back on-line, we get
> all the messages from the system that everything is back up. The maps,
> however, show an inconsistency. The router is green on the network layer
> (the one that shows the router connected to the various segments), but is
> red when you drill down to the individual networks. When I do an ovtopodump
> <router name>, NetView reports the status for all the objects associated
> with the router have a status of Up.
>
> This is kind of a problem since the network support group use the colors to
> determine if there is a problem when they are at their desk. Also, my
> management is a little frustrated since it appears that there is a problem,
> but everything is up and functioning correctly.
>
> Has anyone seen this before?
> Anyone have any suggestions for a way around this?
>
> Thanks,
>
> Patrick
|