Active clients low troughput

[PaAn]

Hi,

I am trying to monitor a bunch of devices which may be behind firewalls and varying network connections. The setup now has the clients in active mode send the data to a zabbix proxy using tls encryption.
Around 800 devices will still work with an required performance (vps) of 65. Much more than this per proxy and the clients will start to lose connections. I can always add proxys but this doesn't scale well.
Does it sound like this setup is horribly misconfigured or is this sort of performance expected with hundreds of active clients.

[vendrusculo]

hi!

have you tried to increase the "trapper" number on proxy? probably it's not big enough to handle 800 machines trying sending data

Thanks,
Leo

[PaAn]

Hi,

Thank you for answering. I have played around a bit with trapper sizes which didn't seem to help much at the time. This was a while ago though and there were a lot more then 800 trying to connect back then and may have simply saturated the network. Will increase the trapper amount a bit and see what it does now.

Wonder how people are load testing zabbix servers. I'm using production devices, which statistics luckily aren't super critical just but nice to have.

[cyber]

Hi,
Around 800 devices will still work with an required performance (vps) of 65. Much more than this per proxy and the clients will start to lose connections.

800 devices and vps only 65? That seems like no load at all. one of my network proxies (99% snmp polls) has vps 726 with 243 hosts.. and has no major issues... A proxy with 1000+ active agents has vps 1100+ and still has default 5 trappers