We avoid using netdata, new relic, Zabbix, etc.. as these are helpful for very very large clusters (1000+ nodes) and sometimes require a extra tweaking to run their 'agents'.. as for netdata, we encountered a similar problem before enhance, when we were managing couple of client servers with plain stack, LEMP or nodejs with pm2 and whatnot, but eventually we removed netdata even though we love its metrics..
TBH, we use as simple as 'glances' monitoring tool along with htop, where glances we use full time! Sometimes going vanilla and as simple as possible helps, unless you absolutely need extra metrics...
