fix: call pg_notification_queue_usage() automatically when the LISTEN channel bug is detected (#4858)

There's a PostgreSQL bug that doesn't let any listener to register in the DB:
https://www.postgresql.org/message-id/flat/CAK98qZ3wZLE-RZJN_Y%2BTFjiTRPPFPBwNBpBi5K5CU8hUHkzDpw%40mail.gmail.com
The only workaround is to advance the async notification queue tail,
which can be done by executing:  "SELECT pg_notification_queue_usage();".
Before we just logged a HINT with this suggestion, but now we call that function directly
and then let the listener to automatically recover.

No automated tests were added here as it would be too complex and this is a PostgreSQL bug. But this was manually tested following the steps on https://github.com/PostgREST/postgrest/pull/4581#issuecomment-3690610592
This commit is contained in:
Laurence Isla
2026-05-04 22:08:06 +05:00
committed by Taimoor Zaeem
parent a62b6de1d0
commit 0502488b70
3 changed files with 19 additions and 14 deletions
+3 -7
View File
@@ -13,7 +13,6 @@ module PostgREST.Observation
, observationMessage
, ObservationHandler
, showOnSingleLine
, isDbListenerBug
) where
import qualified Data.ByteString.Lazy as LBS
@@ -48,7 +47,7 @@ data Observation
| DBListenStart (Maybe ByteString) (Maybe ByteString) Text Text -- host, port, version string, channel
| DBListenFail Text (Either SQL.ConnectionError SomeException)
| DBListenRetry Int
| DBListenBugHint -- https://github.com/PostgREST/postgrest/issues/3147
| DBListenBugCallQueryFix
| DBListenerGotSCacheMsg ByteString
| DBListenerGotConfigMsg ByteString
| DBListenerConnectionCleanupFail SomeException
@@ -122,8 +121,8 @@ observationMessage = \case
either showListenerConnError showListenerException listenErr
DBListenRetry delay ->
"Retrying listening for database notifications in " <> (show delay::Text) <> " seconds..."
DBListenBugHint ->
"HINT: This is likely a bug in the notification queue, try executing the following to solve it: select pg_notification_queue_usage();"
DBListenBugCallQueryFix ->
"This is likely a PostgreSQL bug in the notification queue, executing the following to try to solve it: SELECT pg_notification_queue_usage();"
DBListenerGotSCacheMsg channel ->
"Received a schema cache reload message on the " <> show channel <> " channel"
DBListenerGotConfigMsg channel ->
@@ -188,6 +187,3 @@ observationMessage = \case
showOnSingleLine :: Char -> Text -> Text
showOnSingleLine split txt = T.intercalate " " $ T.filter (/= split) <$> T.lines txt -- the errors from hasql-notifications come intercalated with "\t\n"
isDbListenerBug :: SomeException -> Bool
isDbListenerBug e = "could not access status of transaction" `T.isInfixOf` show e