Why do sensors stop recording data when the probe agent loses connectivity? Shouldn't the probe keep polling and store the data locally until connectivity is restored?

In previous tools that I have used, the probe agent will continue to operate even in the absence of connectivity to the core server (i.e. when VPN connectivity to a separate site is down). The probe still issues SNMP requests to all monitored devices configured for the probe and stores the data and will update as soon as connectivity to the core server is restored. This data is not viewable and no alerts will be sent until connectivity to the core install is resolved, but the data is there once connectivity is back.

The goal is to be able to use your reporting functions to provide network health indicators for monthly % downtime so that we can ensure all SLAs are being met and currently if IPSEC tunnel between core and probe, all devices reflect the % downtime.

Article Comments

In a remote probe setup, whenever the probe–core connection is interrupted, a remote probe will continue to collect data from the network. During a connection loss a buffer stores a maximum of 500,000 sensor results in RAM memory of the remote probe system (up to 50 – 200 MB). This means that for 100 sensors with one minute interval the monitoring results of up to 3 days can be buffered (or 52 minutes for 10,000 sensors with one minute interval). The probe automatically reconnects to the core as soon as it is available again and transmits all monitoring results gathered during the connection loss.

Please note that the computer running the remote probe must not be restarted during an interrupted connection, otherwise monitoring data will be lost.


Jan, 2011 - Permalink

Unfortunately this is not happening...when the connection between our core and probes is interrupted the remote probes may record that data but upon reconnection, all results gathered during the interrupted interval are gone. I have verified that even over a period of weeks this data does not reappear in our graphs or history and I have checked up to 3 times daily because of how important this data is to us.


Jan, 2011 - Permalink

Have you tried recalculating the data cache to see if the data is there, just hadn't been processed properly? What types of sensors are defined on the remote probe? On average, how long is the probe disconnected from the core?


Jan, 2011 - Permalink