DFM managed state

platinum_jem · ‎09-09-2007

Hi Guys,

I wanted to set something like 'maintenence mode' to certain interfaces of managed devices in DFM. I was looking at the 'Managed state' in Detailed Device view and it seems to hint the correct function.

I wanted to place interfaces to maintenence mode so that any traps or events occurred at that port will be ignored and events/alerts wouldn't be sent.

Is the 'Managed state' the function i'm looking for ? The help file states that 'All active events will be moved to cleared state' BUT 'Alerts will remain in active state'.

What is the difference between events and alerts and how does that affects notification ?

Thanks very much for clarification.

Regards,

Sam

Joe Clarke · ‎09-09-2007

Setting an object's managed state to false will prevent new events from being generated for that object. Existing events will be cleared, but the alerts that contain those events will still remain active unless all the events in the given alert are cleared. This sounds like what you are looking for.

An event is an atomic deviation or problem on a device. An alert is a composite of multiple related events.

platinum_jem · ‎09-11-2007

Hi jclarke,

Yes , i have tested it. When i set the managed state to 'false' , no events/alerts are generated when the interface is down.

However , when i change it back to 'True' , immediate if i initiate a up/down on the cables, there are no alerts occurring. In Detailed Device View, it still shows the interface statistics as 'Not Available'. It is until some time before the interface can generate alerts again.

Do i have to do anything else to bring it back to normal?

An additional question. How did Ciscoworks know when interfaces goes down on a Managed Device ? Is it by regular polling or the Managed Device need to send traps to Ciscoworks when interface goes down ?

Regards,

Sam

Joe Clarke · ‎09-11-2007

Whenever you change the managed state to true, you also have to go to DFM > Configuration > Polling and Thresholds > Apply Changes before the object will actually become managed again.

DFM will detect an interface is down by periodic polling.

platinum_jem · ‎09-12-2007

Hi,

Is there any other way to 'un-monitor' a port other than changing the Managed state to false ? It seems strange that everytime you do that , u need to 'Apply Changes' to make it work again.

Which polling interval defines it ? If i set the interval to be 240secs , then DFM would have to wait 240secs before detecting the downed interface ?? I always thought it was snmp trap sent to Ciscoworks..

Joe Clarke · ‎09-12-2007

Setting an object's managed state to false does not require one to apply changes. The object will immediately become unmanaged.

Yes, the default is 240 seconds. DFM will also process linkUp and linkDown traps, but it uses those to determine if a link is flapping as opposed to simply down. See http://www.cisco.com/en/US/docs/net_mgmt/ciscoworks_device_fault_manager/2.0_IDU_2.0.6/user/guide/TrapFwd.html for more on what traps DFM handles.