Feat: Healthcheck endpoints in deploy #37
Conversation
| // after the HTTP server starts. | ||
| private val livenessEndpoint: PublicEndpoint[Unit, Nothing, Unit, Any] = | ||
| infallibleEndpoint.in("health").in("live").get | ||
| private val readinessEndpoint: PublicEndpoint[Unit, Nothing, Unit, Any] = |
There was a problem hiding this comment.
Do we need 3 endpoints if all of them do exactly same thing for now?
(I don't have a strong preference) but I think it would make more sense to add more endpoints if and when we need them (we can start with just one)
There was a problem hiding this comment.
one advantage of 3 endpoints is that external users can hook up their startup/readiness/liveness probes to the endpoins and if we want to have different implementation in the future, they just upgrade, no need to change their helmrelease/whatever deployment.
There was a problem hiding this comment.
Metrics server needs these too but can be a separate PR.
I don't remember how the startup is done now, but there we can start the http server before the "__consumer_offsets" topic is read until the end (we can still serve topic offsets). I don't have a strong opinion though, let's explore the options and discuss what's best.
j-hall-mwam
force-pushed
the
scalafmt
branch
from
5e9802c to
0f14847
Compare
j-hall-mwam
force-pushed
the
healthcheck_endpoints
branch
from
53f1801 to
161444f
Compare
j-hall-mwam
force-pushed
the
scalafmt
branch
4 times, most recently
from
234977a to
0b5cf76
Compare
j-hall-mwam
force-pushed
the
healthcheck_endpoints
branch
from
161444f to
a1d38d4
Compare
Signed-off-by: Jordan Hall <[email protected]>
j-hall-mwam
force-pushed
the
healthcheck_endpoints
branch
from
a1d38d4 to
8e705a2
Compare
j-hall-mwam
changed the title
Add the following HTTP healthcheck endpoints to the deploy service:
/health/live/health/ready/health/startupCurrently, these always return status code 200 (OK), as setup is always completed before HTTP server comes online.