Polling Timeout Warning
SolarWinds recommends monitoring a maximum of 3,000 devices with a single UDT poller. If many of your devices require monitoring at both Layer 2 and Layer 3 of the network stack, the maximum device limit that a single UDT poller can service within the polling interval (by default, 30 minutes) may be lower.
For example, if you use the default polling interval (30 minutes), you are managing 2,000 devices on a server with the recommended minimum hardware widgets, and 1,000 devices require polling at both L2 and L3, the polling cycle may complete before 1,000 jobs are even started. As a result, those unstarted jobs are queued for the next polling interval. However, in the meantime, the jobs currently running, having started, continue to be processed until finished, which may be sometime after 30 minutes—let’s say 35 minutes.
35 minutes after the previous polling cycle, the next cycle begins and the jobs queued from the previous cycle are processed. And this staggered pattern of accumulating data is likely to become the norm, potentially producing issues with the timeliness of data seen in Web Console widgets and reports.
There are three options:
- Increase server hardware components so that the existing polling workload finishes within the set interval.
- Install an additional UDT poller on a separate server machine, and distribute work across the two pollers.
- Increase the polling intervals, with the risk that the data gathered for devices polled early in the cycle might become stale by the time the cycle finishes.