This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Topology is wrong

In general the topology of my network that NPM has discovered is quite accurate... but there is one switch that has the topology completely wrong and I don't know why.

Specially, I have a HP 3500yl connected with a single link/vlan to a HP 5400 core switch. Several other switches are also linked to the 5400 core, and all switches have an interface on the same vlan for management purposes But Orion shows the 3500yl as connecting DIRECTLY to ALL of the switches connected to the 5400, not JUST the 5400 itself, which is what the physical/logical reality is.

I am trying to determine if Orion is just reporting things inaccurately, or if I actually have something mis-configured? One thing that sets the 3500yl apart from the "other" switches (besides the core) is that it is an L3 switch, but does not even have ip routing enabled... could this have something to do with Orion reporting an incorrect topology?

  • Can you please use list resources on both your HP switches and make sure there are L2 topology pollers available and selected on at least one of them? Connections created using ARP table data are less accurate than Bridge-MIB or Q-Bridge-MIB connections. If at least one of those two HPs has L2 poller assigned, please open a support case, so we can take a look.

    Note:

    We've made significant improvements in topology polling and calculation in upcoming release of NPM. Orion NPM 10.3 BETA 3 was released few days ago, would it be possible for you to join the BETA/RC program?

  • Thanks for the reply... Both switches have BOTH Topology Layer 2 and Topology Layer 3 selected (I'm unclear when to select one over the other, so pretty much all of my switches have both selected... is that bad?) Anyway, the one thing that separates this "odd" switch from the rest is that it IS a layer 3 switch, but layer3 functions are disabled. I went ahead and disabled L3 (leaving L2 selected on both the 3500 and 5400 core) but that didn't change the topology, either. Does the topology stay "fixed" once it's learned, and doesn't change when configurations change, or is it dynamic?

    Bottom line is that I still have this one weird L3 (but not doing L3) switch thinking it's directly connected to five other switches.

    What's involved in joining the beta?

  • Orion is cleaning topology data tables during regular database maintenance. It deletes topology data related to nodes which were deleted, or nodes which no longer have topology pollers assigned. So it takes some time, but the incorrect link should get removed at some point. If you eventually decide to open a support ticket for this issue, please post the case number here, so I can take a look.

    Please take a look on NPM 10.3 Beta 3 is now available! forum on thwack to get more details about BETA and RC programs.

    Thanks!

    Marek Zizlavsky

    SolarWinds Developer

  • What is the answer to pufferdude's question? I assume it's more accurate to disable L3 whether or not our switches are Layer 3?

  • Layer 3 topology poller is collecting ARP table data, which Orion may use for building topology connections. ARP table may help to add some missing connections between nodes, but sometimes it just adds a false ones. Information from ARP table is not very reliable, because it contains both direct and indirect connections.