Restore one-minute retries for healthcheck delivery

CONTRIBUTING.md

@@ -64,21 +64,7 @@ The server will load environment variables from the `.env` file in the project r
 
 ### Healthchecks Reporting
 
-Set `HEALTHCHECKS_URL` to a Healthchecks success ping URL to enable operational
-reports. After a one-minute startup allowance, the server sends a POST every
-five minutes, including a short diagnostic body. Configure the check with a
-five-minute period and ten minutes of grace.
-
-A successful database query and an operational job executor produce a success
-ping. An initializing, restarting, or livelock-paused executor, or a failed
-database query, produces a `/fail` ping. Idle and busy executors are both healthy;
-an individual failed package job does not mark the service unhealthy. Reports
-time out after ten seconds and retry on the next scheduled report indefinitely.
-
-This is an internal operational check, not a probe of public DNS, TLS, or nginx,
-and does not detect every possible stalled job or external dependency failure.
-`/api/v1/status` remains a simple HTTP liveness check. Ping URLs are secrets: do
-not commit them or expose them in logs.
+Set `HEALTHCHECKS_URL` to a Healthchecks success ping URL to enable operational reports. After a one-minute startup allowance, the server sends a POST every five minutes, including a short diagnostic body. This expects a simple schedule with five-minute internals and ten minutes of grace.
 
 ## Quick Start: Running Integration Tests
 

app/src/App/Server/Healthcheck.purs

@@ -3,7 +3,6 @@ module Registry.App.Server.Healthcheck (run, report) where
 import Registry.App.Prelude
 
 import Control.Parallel as Parallel
-import Data.DateTime (diff)
 import Data.Time.Duration (Milliseconds(..))
 import Effect.Aff as Aff
 import Effect.Class.Console as Console
@@ -12,15 +11,14 @@ import Fetch as Fetch
 import Registry.App.Effect.Db as Db
 import Registry.App.Server.Env (ServerEnv, runEffects)
 import Registry.App.Server.Env as Env
+import Registry.Foreign.Console as Console.Foreign
 
--- | Give startup one minute, then report on a five-minute cadence. Failures to
--- | deliver a ping never stop the reporter; the next scheduled ping retries.
+-- | Give startup one minute, then wait five minutes after a delivered report.
+-- | Failed deliveries retry after one minute, indefinitely.
 run :: ServerEnv -> String -> Aff Unit
 run env url = Aff.delay (Milliseconds 60_000.0) *> loop
   where
-  interval = Milliseconds 300_000.0
   loop = do
-    start <- nowUTC
     database <- runEffects env $ void Db.selectNextPublishJob
     executor <- liftEffect $ Ref.read env.executorStatus
     let
@@ -32,11 +30,12 @@ run env url = Aff.delay (Milliseconds 60_000.0) *> loop
           Env.Paused -> Left "Executor paused after repeated job resets"
           Env.Restarting -> Left "Executor restarting after unexpected exit; see server logs"
     result <- Aff.attempt $ report url health
-    for_ (either Just (const Nothing) result) \_ ->
-      Console.warn "Healthcheck report failed; retrying next scheduled report"
-    end <- nowUTC
-    let Milliseconds elapsed = end `diff` start
-    Aff.delay $ Milliseconds $ max 0.0 (unwrap interval - elapsed)
+    case result of
+      Left error -> do
+        Console.warn "Healthcheck report failed; retrying in one minute:"
+        liftEffect $ Console.Foreign.warnError error
+        Aff.delay $ Milliseconds 60_000.0
+      Right _ -> Aff.delay $ Milliseconds 300_000.0
     loop
 
 -- | Bound requests and use abortable fetch so an unreachable monitoring service

foreign/src/Foreign/Console.js

@@ -0,0 +1,3 @@
+import { inspect } from "node:util";
+
+export const warnError = (error) => () => console.warn(inspect(error, { depth: null }));

foreign/src/Foreign/Console.purs

@@ -0,0 +1,8 @@
+module Registry.Foreign.Console (warnError) where
+
+import Prelude
+
+import Effect (Effect)
+import Effect.Exception (Error)
+
+foreign import warnError :: Error -> Effect Unit