Hi All
I canned my previous attempt at distributed polling due to the high quantity of data between poller and server as the pollers were located at remote sites with limited data (LTE connections mostly), so have been running 2 separate servers (one with ±1200 devices and another with 1300). The one crashed the other day and I’ve decided to rather combine the two, but with “local” distributed polling (due to the quantity of devices).
So I’ve configured the main server (8 x cores with 8Gb RAM) to do everything including polling and then fired up a 2nd VM (Poller 1, 6 x cores with 8Gb RAM) to help with it’s current load (it was starting to struggle a tad anyway) and then a 3rd VM (Poller 2, 8 x cores with 16Gb RAM) to carry the load for the server that crashed.
I configured another poller group so I can isolate the one poller but no matter where I set the Group Served (config, lnms command, Poller settings etc), it eventually defaults back to 0. How do I make it stick?
Oh and the same goes for Workers, I managed to change it somewhere from 20 to 25 but now it refuses to go back to 20 again.
Also why are the “Devices Actioned” per poller quantities nearly 5 times the actual device count?
And while it’s running with 3 separate pollers with only one server’s load of devices I thought it would just be idling through everything, but it still seems to be running at max. How is that possible?
Validate looks exactly the same for all 3 … validate - LibreNMS
I’ve run through the basics on the Performance page (bar DNS, max OIDs, SNMP repeaters as it doesn’t seem to do much and opcache).
Thanks,
R