feat: minimal health check (#2092)

CHANGELOG.md

@@ -7,6 +7,9 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 
 ### Added
 
+ - #1933, Add a minimal health check endpoint on an admin port at the `<host>:<admin_server_port>/health` endpoint - @steve-chavez
+   + For enabling this, the `admin-server-port` config must be set explictly
+
 ### Fixed
 
  - #2020, Execute deferred constraint triggers when using `Prefer: tx=rollback` - @wolfgangwalther

src/PostgREST/App.hs

@@ -30,9 +30,10 @@ import qualified Data.ByteString.Char8           as BS
 import qualified Data.ByteString.Lazy            as LBS
 import qualified Data.HashMap.Strict             as M
 import qualified Data.Set                        as S
-import qualified Hasql.DynamicStatements.Snippet as SQL
+import qualified Hasql.DynamicStatements.Snippet as SQL (Snippet)
 import qualified Hasql.Pool                      as SQL
-import qualified Hasql.Transaction               as SQL
+import qualified Hasql.Session                   as SQL (sql)
+import qualified Hasql.Transaction               as SQL hiding (sql)
 import qualified Hasql.Transaction.Sessions      as SQL
 import qualified Network.HTTP.Types.Header       as HTTP
 import qualified Network.HTTP.Types.Status       as HTTP
@@ -113,6 +114,11 @@ run installHandlers maybeRunWithSocket appState = do
   when configDbChannelEnabled $ listener appState
 
   let app = postgrest configLogLevel appState (connectionWorker appState)
+      adminApp = postgrestAdmin appState configDbChannelEnabled
+
+  whenJust configAdminServerPort $ \adminPort -> do
+    AppState.logWithZTime appState $ "Admin server listening on port " <> show adminPort
+    void . forkIO $ Warp.runSettings (serverSettings conf & setPort adminPort) adminApp
 
   case configServerUnixSocket of
     Just socket ->
@@ -127,6 +133,9 @@ run installHandlers maybeRunWithSocket appState = do
       do
         AppState.logWithZTime appState $ "Listening on port " <> show configServerPort
         Warp.runSettings (serverSettings conf) app
+  where
+    whenJust :: Applicative m => Maybe a -> (a -> m ()) -> m ()
+    whenJust mg f = maybe (pure ()) f mg
 
 serverSettings :: AppConfig -> Warp.Settings
 serverSettings AppConfig{..} =
@@ -135,6 +144,19 @@ serverSettings AppConfig{..} =
     & setPort configServerPort
     & setServerName ("postgrest/" <> prettyVersion)
 
+-- | PostgREST admin application
+postgrestAdmin :: AppState.AppState -> Bool -> Wai.Application
+postgrestAdmin appState configDbChannelEnabled req respond  =
+  case Wai.pathInfo req of
+    ["health"] ->
+      if configDbChannelEnabled then do
+        listenerOn <- AppState.getIsListenerOn appState
+        respond $ Wai.responseLBS (if listenerOn then HTTP.status200 else HTTP.status503) [] mempty
+      else do
+        result <- SQL.use (AppState.getPool appState) $ SQL.sql "SELECT 1"
+        respond $ Wai.responseLBS (if isRight result then HTTP.status200 else HTTP.status503) [] mempty
+    _ -> respond $ Wai.responseLBS HTTP.status404 [] mempty
+
 -- | PostgREST application
 postgrest :: LogLevel -> AppState.AppState -> IO () -> Wai.Application
 postgrest logLev appState connWorker =

src/PostgREST/AppState.hs

@@ -4,6 +4,7 @@ module PostgREST.AppState
   ( AppState
   , getConfig
   , getDbStructure
+  , getIsListenerOn
   , getIsWorkerOn
   , getJsonDbS
   , getMainThreadId
@@ -16,6 +17,7 @@ module PostgREST.AppState
   , logWithZTime
   , putConfig
   , putDbStructure
+  , putIsListenerOn
   , putIsWorkerOn
   , putJsonDbS
   , putPgVersion
@@ -53,6 +55,8 @@ data AppState = AppState
   , stateIsWorkerOn   :: IORef Bool
   -- | Binary semaphore used to sync the listener(NOTIFY reload) with the connectionWorker.
   , stateListener     :: MVar ()
+  -- | State of the LISTEN channel, used for health checks
+  , stateIsListenerOn :: IORef Bool
   -- | Config that can change at runtime
   , stateConf         :: IORef AppConfig
   -- | Time used for verifying JWT expiration
@@ -78,6 +82,7 @@ initWithPool newPool conf =
     <*> newIORef mempty
     <*> newIORef False
     <*> newEmptyMVar
+    <*> newIORef False
     <*> newIORef conf
     <*> mkAutoUpdate defaultUpdateSettings { updateAction = getCurrentTime }
     <*> mkAutoUpdate defaultUpdateSettings { updateAction = getZonedTime }
@@ -153,3 +158,9 @@ waitListener = takeMVar . stateListener
 -- the connectionWorker is the only mvar producer.
 signalListener :: AppState -> IO ()
 signalListener appState = void $ tryPutMVar (stateListener appState) ()
+
+getIsListenerOn :: AppState -> IO Bool
+getIsListenerOn = readIORef . stateIsListenerOn
+
+putIsListenerOn :: AppState -> Bool -> IO ()
+putIsListenerOn = atomicWriteIORef . stateIsListenerOn

src/PostgREST/CLI.hs

@@ -199,6 +199,9 @@ exampleConfigFile =
       |## when none is provided, 660 is applied by default
       |# server-unix-socket-mode = "660"
       |
+      |## admin server for health checks, it's disabled by default unless a port is specified
+      |# admin-server-port = 3001
+      |
       |## determine if the OpenAPI output should follow or ignore role privileges or be disabled entirely
       |## admitted values: follow-privileges, ignore-privileges, disabled
       |openapi-mode = "follow-privileges"

src/PostgREST/Config.hs

@@ -93,6 +93,7 @@ data AppConfig = AppConfig
   , configServerPort            :: Int
   , configServerUnixSocket      :: Maybe FilePath
   , configServerUnixSocketMode  :: FileMode
+  , configAdminServerPort       :: Maybe Int
   }
 
 data LogLevel = LogCrit | LogError | LogWarn | LogInfo
@@ -147,6 +148,7 @@ toText conf =
       ,("server-port",                   show . configServerPort)
       ,("server-unix-socket",        q . maybe mempty T.pack . configServerUnixSocket)
       ,("server-unix-socket-mode",   q . T.pack . showSocketMode)
+      ,("admin-server-port",             maybe "\"\"" show . configAdminServerPort)
       ]
 
     -- quote all app.settings
@@ -242,6 +244,7 @@ parser optPath env dbSettings =
     <*> (fromMaybe 3000 <$> optInt "server-port")
     <*> (fmap T.unpack <$> optString "server-unix-socket")
     <*> parseSocketFileMode "server-unix-socket-mode"
+    <*> optInt "admin-server-port"
   where
     parseAppSettings :: C.Key -> C.Parser C.Config [(Text, Text)]
     parseAppSettings key = addFromEnv . fmap (fmap coerceText) <$> C.subassocs key C.value

src/PostgREST/Workers.hs

@@ -200,6 +200,7 @@ listener appState = do
     case dbOrError of
       Right db -> do
         AppState.logWithZTime appState $ "Listening for notifications on the " <> dbChannel <> " channel"
+        AppState.putIsListenerOn appState True
         SQL.listen db $ SQL.toPgIdentifier dbChannel
         SQL.waitForNotifications handleNotification db
       _ ->
@@ -208,6 +209,7 @@ listener appState = do
     handleFinally dbChannel _ = do
       -- if the thread dies, we try to recover
       AppState.logWithZTime appState $ "Retrying listening for notifications on the " <> dbChannel <> " channel.."
+      AppState.putIsListenerOn appState False
       -- assume the pool connection was also lost, call the connection worker
       connectionWorker appState
       -- retry the listener

test/SpecHelper.hs

@@ -106,6 +106,7 @@ _baseCfg = let secret = Just $ encodeUtf8 "reallyreallyreallyreallyverysafe" in
   , configServerUnixSocketMode  = 432
   , configDbTxAllowOverride     = True
   , configDbTxRollbackAll       = True
+  , configAdminServerPort       = Nothing
   }
 
 testCfg :: Text -> AppConfig

test/io-tests/configs/expected/aliases.config

@@ -25,3 +25,4 @@ server-host = "!4"
 server-port = 3000
 server-unix-socket = ""
 server-unix-socket-mode = "660"
+admin-server-port = ""

test/io-tests/configs/expected/boolean-numeric.config

@@ -25,3 +25,4 @@ server-host = "!4"
 server-port = 3000
 server-unix-socket = ""
 server-unix-socket-mode = "660"
+admin-server-port = ""

test/io-tests/configs/expected/boolean-string.config

@@ -25,3 +25,4 @@ server-host = "!4"
 server-port = 3000
 server-unix-socket = ""
 server-unix-socket-mode = "660"
+admin-server-port = ""

test/io-tests/configs/expected/defaults.config

@@ -25,3 +25,4 @@ server-host = "!4"
 server-port = 3000
 server-unix-socket = ""
 server-unix-socket-mode = "660"
+admin-server-port = ""

test/io-tests/configs/expected/no-defaults-with-db-other-authenticator.config

@@ -25,5 +25,6 @@ server-host = "0.0.0.0"
 server-port = 80
 server-unix-socket = "/tmp/pgrst_io_test.sock"
 server-unix-socket-mode = "777"
+admin-server-port = 3001
 app.settings.test = "test"
 app.settings.test2 = "test"

test/io-tests/configs/expected/no-defaults-with-db.config

@@ -25,5 +25,6 @@ server-host = "0.0.0.0"
 server-port = 80
 server-unix-socket = "/tmp/pgrst_io_test.sock"
 server-unix-socket-mode = "777"
+admin-server-port = 3001
 app.settings.test = "test"
 app.settings.test2 = "test"

test/io-tests/configs/expected/no-defaults.config

@@ -25,5 +25,6 @@ server-host = "0.0.0.0"
 server-port = 80
 server-unix-socket = "/tmp/pgrst_io_test.sock"
 server-unix-socket-mode = "777"
+admin-server-port = 3001
 app.settings.test = "test"
 app.settings.test2 = "test"

test/io-tests/configs/expected/types.config

@@ -25,4 +25,5 @@ server-host = "!4"
 server-port = 3000
 server-unix-socket = ""
 server-unix-socket-mode = "660"
+admin-server-port = ""
 app.settings.test = "Bool False"

test/io-tests/configs/no-defaults-env.yaml

@@ -28,3 +28,4 @@ PGRST_SERVER_HOST: 0.0.0.0
 PGRST_SERVER_PORT: 80
 PGRST_SERVER_UNIX_SOCKET: /tmp/pgrst_io_test.sock
 PGRST_SERVER_UNIX_SOCKET_MODE: 777
+PGRST_ADMIN_SERVER_PORT: 3001

test/io-tests/configs/no-defaults.config

@@ -25,5 +25,6 @@ server-host = "0.0.0.0"
 server-port = 80
 server-unix-socket = "/tmp/pgrst_io_test.sock"
 server-unix-socket-mode = "777"
+admin-server-port = 3001
 app.settings.test = "test"
 app.settings.test2 = "test"

test/io-tests/test_io.py

@@ -729,3 +729,44 @@ def test_db_prepared_statements_disable(defaultenv):
     with run(env=env) as postgrest:
         response = postgrest.session.post("/rpc/uses_prepared_statements")
         assert response.text == "false"
+
+
+def test_admin_healthy_w_channel(defaultenv):
+    "Should get a success response from the admin server health endpoint when the LISTEN channel is enabled"
+
+    env = {
+        **defaultenv,
+        "PGRST_ADMIN_SERVER_PORT": "3001",
+        "PGRST_DB_CHANNEL_ENABLED": "true",
+    }
+
+    with run(env=env) as postgrest:
+        response = requests.get(f"http://localhost:{env['PGRST_ADMIN_SERVER_PORT']}/health")
+        assert response.status_code == 200
+
+
+def test_admin_healthy_wo_channel(defaultenv):
+    "Should get a success response from the admin server health endpoint when the LISTEN channel is disabled"
+
+    env = {
+        **defaultenv,
+        "PGRST_ADMIN_SERVER_PORT": "3001",
+        "PGRST_DB_CHANNEL_ENABLED": "false",
+    }
+
+    with run(env=env) as postgrest:
+        response = requests.get(f"http://localhost:{env['PGRST_ADMIN_SERVER_PORT']}/health")
+        assert response.status_code == 200
+
+
+def test_admin_not_found(defaultenv):
+    "Should get a not found from the admin server"
+
+    env = {
+        **defaultenv,
+        "PGRST_ADMIN_SERVER_PORT": "3001",
+    }
+
+    with run(env=env) as postgrest:
+        response = requests.get(f"http://localhost:{env['PGRST_ADMIN_SERVER_PORT']}/notfound")
+        assert response.status_code == 404