Forum Discussion

Stuart_Taylor's avatar
11 years ago

Monitoring datapoints that are hours apart

Has anyone tried to monitor a source which only produces data every few hours. Typically the system in questions processes 1000 items in a burst and this happens randomly every few hours with no activity in between. We\'ve been asked to monitor this for when it stops processing. We don\'t want to set the datasource to only poll every few hours as that will lose accurate flow information.

Anyone got any bright ideas?

  • Hi Stuart,rnIf your device only reports metrics every few hours, then we will polling for data more frequently than what your device is reporting. But, correspondingly, we will still show these changes that represent an influx of data in your graphs as they come in every few hours.rnBut, from what you are telling me, it sounds like the question of how alerting would work is more of a concern in this case. I think what might work in your case is to alert on a positive delta threshold, and by making wise use of your Alert Transition Trigger intervals.rn rnThe boolean operators that we support for alert thresholds are detailed at the bottom of this page:rnhttp://help.logicmonitor.com/using/datasource/logicmonitor-scripting-language-lms/rn rnApart from that, there might be a SNMP trap event that gets paired with these sporadic data updates, and this might be a rare case where you may actually want to monitor the pushed SNMP trap events via an eventsource:rn rnFor more information:rnhttp://help.logicmonitor.com/using/event-sources/snmp-traps-monitoring/rnhttp://blog.logicmonitor.com/2013/12/08/why-relying-on-snmp-traps-is-a-bad-idea/rn rnKind regard,rn rnJoseph Skup.rnLogicMonitor Support