OverviewUsers may sometimes see some gaps in the monitoring graphs for Interfaces/link utilization. Just like the image below
CauseThis can be caused by
1. The Exinda is in bypass for the time of the gap
2. The network was down (no traffic passing through the interfaces) for the time of the gap
3. The process statsd responsible for displaying this information was not running for the time of the gap
WorkaroundIf the reason for the gaps is #1 or #2, check the network, the cables on the Exinda or even the logs on the Exinda to see if the device was up and running during that time.
If the network was fine, check if other graphs like applications graphs has data for the same time. If it does, then it is most probably that the process statsd was not connected to the monitoring module. Hence it could not display information for the missing time. This generally happens when the process mgmtd (for management interface) and statsd (for stats display) are busy and can't keep up with the monitoring module on the Exinda. There also might be also see auto-generated sysdumps or snapshots for statsd if the process crashed (found under Configuration > System > Diagnostics, "System Diagnostics" tab).
The above can happen if the device is busy and the memory utilization is very high. The device has a lot of network objects or policies or virtual circuits which are enabled for monitoring. Due to all this monitoring it is unable to keep up with the amount if information which needs processing and hence ends up crashing
ResolutionIf the case is the latter and statsd has crashed due to high resource usage:
- Disable subnet monitoring for network objects that are not critical for monitoring
- Try to consolidate the optimizer config such that the number of VC's or policies are not too high in number.
- Run scheduled reports during after hours, so that if the process statsd is trying to generate a report, it gets enough amount of RAM to work with.