Verified that the "dummy" message is successfully logged to New Relic when New Relic credentials are present.
Learned that the fluentbit -P flag specifies the port for the optional fluentbit HTTP server, not the port for fluentbit to listen on, so removed that from the startup script.
Drafted an INPUT section for syslogs. fluentd and fluentbit syntax diverge on this one.
Working on:
Proxy settings are not working in cloud.gov environment. I'm still noodling on this one. Fluentbit uses the HTTP_PROXY (or http_proxy) environment variable and only accepts "http" (not "https") as the protocol, but their docs claim that this works for HTTPS as well. (Current theory: HTTPS is supported, and the problem is elsewhere.)
I was unable to verify the INPUT configuration using an app deployed in my sandbox environment because I could not set up a tcp route to that app. A log drain service pointed at an https route made http requests, which fluentbit was unable to parse with the syslog-rfc5424 parser (understandably!)
Either cloud.gov does not allow users to create tcp routes, or ...
I can create a tcp route, but first I must have a domain for it (guessing that all *.app.cloud.gov routes rely on ALB).
Progress so far:
Working on: