Although I've successfully used collectd in the past, I would choose telegraf over it now. This is especially true if I'm sending data to InfluxDB,s ince it and telegraf are primarily developed by InfluxData.
Both telegraf and collectd are mature programs with a large base of well-documented plugins. However, telegraf development is more active, and I find its golang codebase much more approachable than the C codebase of collectd. Obviously, YMMV.
In general though, I wouldn't advise adopting the entire TICK stack. Chronograf is inferior to Grafana. Kapacitor has a super-steep learning curve and is being deprecated by InfluxData. TIG (telegraf, influxdb, grafana) can be a good choice. Prometheus plus Grafana is worth evaluating as well; there's arguably more momentum in that ecosystem than in InflxuDB's.
Both telegraf and collectd are mature programs with a large base of well-documented plugins. However, telegraf development is more active, and I find its golang codebase much more approachable than the C codebase of collectd. Obviously, YMMV.
In general though, I wouldn't advise adopting the entire TICK stack. Chronograf is inferior to Grafana. Kapacitor has a super-steep learning curve and is being deprecated by InfluxData. TIG (telegraf, influxdb, grafana) can be a good choice. Prometheus plus Grafana is worth evaluating as well; there's arguably more momentum in that ecosystem than in InflxuDB's.