Currently, Telegraf only gets "*.notice" logs as an input
The outputs that are configurable are well supported for ingesting logs as well as the metrics that telegraf currently imports.
By default, the 50-telegraf.conf syslog configuration should pull in more logs, and avoid using rsyslog as the interface, so that Telegraf as a central process can be the single point of log/metric egress from n inputs to m outputs.