To reiterate what @electrocret said, there is a list of sanity checking here: Performance - LibreNMS Docs
I also had particular success tuning SNMP Max Repeaters when I was monitoring things over variable latency, which I over-explain here: Debugging graph spikes from high latency links - #6 by rhinoau
My issues were causing discovery interruptions and scambling my data and spiking traffic, I was still completing polling within the window, though the max repeaters tuning dramatically reduced my polling time.
Poller performance graphs are a good place to start to see what is taking up the time and maybe blowing out polling past the 5 minute (default) window:
The log tab on that same page shows each device total time as it comes in too so you can start to target the slower devices.