IP SLA reaction and its state

dmitry · ‎02-08-2009

Hi,

Have a question if IP SLA feature maintains a state of the reaction and if yes, for how long and if it is configurable, for instance when a packet loss is detected by SLA and a reaction trap is sent, then a month later the packet loss problem went away, however SLA will not send recovery (falling) trap. Apparently, there is a state for 24 hours (packet loss is detected at 12:01AM (rising trap is sent), 11:59PM no more packet loss and there will be falling trap sent)

Thanks

peterlmyers · ‎02-09-2009

Hi Dmitry,

I have recently come across this same issue and couldn't find any information on fixing it.

The solution we went for was to leave the IP SLA running (life forever) and actively SNMP poll the box rather than rely on the traps. You can poll for the latest SLA state (success/failure) as well as values from the SLA itself such as thresholds or RTTs.

Traps would be nice but you do get more flexibility when polling.

Cheers

dmitry · ‎02-09-2009

Hi Peter,

Thanks for your response, good idea to use the polling to get the state change. I'm trying to hook this IP SLA to a Nagios with passive checks and it (Nagios) does have a solution for exactly that type of situation (noramly relies on passive notifications (traps) but can also actively verify the state if it thinks that this state is no longer valid)