NetXMS Support Forum

I'm looking for a possibility to have unmanaged switches as managed nodes in NetXMS.

Example:
NetXMS --- SwitchA --- SwitchB --- DeviceX
(NetXMS = Monitoring-Server, SwitchA = managed Switch, SwitchB = unmanaged Switch, DeviceX = some NetXMS node)

So I could use the Status of the switch port on SwitchA (on which SwitchB is connected) to determine the status of SwitchB. I defined for SwitchB a node in NetXMS with a DCI which reads the interface status from SwitchA. But so I only have a normal DCI on SwitchB and no 'Status' (with SYS_NODE-DOWN and UP).
I made some experiments with the 'This is address of remote management node' switch. But no luck.
Has anybody a good solution for such a situation? Thanks in advance for all suggestions.

Without changes in NetXMS itself the only way to generate node down/up events for node without IP address is to use script in event processing policy which will use PostEvent to send SYS_NODE_DOWN or SYS_NODE_UP event when threshold is violated on interface status DCI.

I'm now thinking that it could be worth to add special functionality for such devices - probably by adding option to manually specify peer interface on managed node. Then status poll will check interface status and if it is down mark unmanageable node as down.

Best regards,
Victor

Thanks for your fast reply.
I tried your first suggestion but the node status only changes between 'critical' and 'unknown', no 'normal' when UP.
I also tried the unmanaged switch with a unexisting IP address (Polling: all=off beside status-polling and data collection), and a threshold on the DCI generating directly SYS_NODE_DOWN and SYS_NODE_UP, so I get UP and DOWN events but the node status changed also between 'unkown' and 'critical' but never 'normal'.
Do you know why this status 'unknown'?


But the new feature in your second passage would be much better and could help us a lot.

This is because if there are no interfaces to poll NetXMS server uses only alarms to determine node status. You can get "normal" status if you will generate "normal" severity alarm from SYS_NODE_UP event. This is not elegant solution, I think I will move forward with implementing second option in near future - it should not take much effort to implement.

Best regards,
Victor

I 'solved' the unknown state by creating a pseudo interface on the node with excpected state = ignore and activating ICMP-pings for status polling. I know, absolutely not very elegant, more like bricolage :-)

Please take the second option as a feature request! I don't know what would be better: take the interface status of an other node for status or more generally a free configurable DCI (so you can check not only interface status of another node).

for a more elegant methode (in NetXMS V1.2.17) see this thread:
https://www.netxms.org/forum/configuration/node-with-question-mark-on-the-icon-of-object-browser/