docs/onboarding.md is 520 lines on bringing your own domain. Its author has
read it many times. Nobody else has executed it.
This is the most valuable item in this list and the least like work. The standing
goal is an engine where cluster knowledge lives in config and the signal surface
rather than in Go. Every piece of evidence for that today is a test I wrote,
passing on a rig I built. Someone following the doc against any other cluster
produces the first independent evidence either way, and the bug list falls out of
the attempt for free.
The offline half needs no cluster at all:
go test ./test/onboarding -v
That drives all five beats over a domain authored entirely in config, named
acme on purpose so the day onboarding needs a domain-specific Go discriminator,
it appears in test/onboarding/ and gets caught.
Done when: someone who didn't write it has followed it end to end against a
cluster that isn't thump-test, and filed what broke.
docs/onboarding.mdis 520 lines on bringing your own domain. Its author hasread it many times. Nobody else has executed it.
This is the most valuable item in this list and the least like work. The standing
goal is an engine where cluster knowledge lives in config and the signal surface
rather than in Go. Every piece of evidence for that today is a test I wrote,
passing on a rig I built. Someone following the doc against any other cluster
produces the first independent evidence either way, and the bug list falls out of
the attempt for free.
The offline half needs no cluster at all:
go test ./test/onboarding -vThat drives all five beats over a domain authored entirely in config, named
acmeon purpose so the day onboarding needs a domain-specific Go discriminator,it appears in
test/onboarding/and gets caught.Done when: someone who didn't write it has followed it end to end against a
cluster that isn't
thump-test, and filed what broke.