From 680d879d034adffa4df1dd99620ad819061d73ce Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Laurent=20Gourv=C3=A9nec?= <lgourvenec@cadoles.com>
Date: Thu, 21 Aug 2025 14:17:25 +0200
Subject: [PATCH] feat(hydra-cleaner): clean authentication_session table

---
 .../hydra-cleaner/files/hydra-cleaner.sh      | 73 ++++++++++++++++++-
 1 file changed, 72 insertions(+), 1 deletion(-)

diff --git a/components/hydra-cleaner/files/hydra-cleaner.sh b/components/hydra-cleaner/files/hydra-cleaner.sh
index da52fe7..889f522 100644
--- a/components/hydra-cleaner/files/hydra-cleaner.sh
+++ b/components/hydra-cleaner/files/hydra-cleaner.sh
@@ -16,10 +16,13 @@ set -o nounset
 
 # -> delete "cascade" on table "flow" cleans access, code, oidc, pkce and refresh tables.
 
+# For table authentication_session, a simple delete on this table is enough.
+# Delete all rows without reference from table flow (foreign key)
 
 DSN="${DSN:-postgresql://${HYDRA_DATABASE_USER}:${HYDRA_DATABASE_PASSWORD}@${HYDRA_DATABASE_SERVICE_NAME}:${HYDRA_DATABASE_SERVICE_PORT:-5432}/hydra?sslmode=disable}"
 RETENTION_HOURS="${RETENTION_HOURS:-48}"
 BATCH_SIZE="${BATCH_SIZE:-50}"
+BATCH_SIZE_ORIG="${BATCH_SIZE}"
 LIMIT="${LIMIT:-1000}"
 BEFORE_DATE="$(date +'%Y-%m-%d %H:%M:%S' --date=@$(($(date +%s) - RETENTION_HOURS * 3600)))"
 
@@ -66,6 +69,7 @@ where table_schema = 'public'
 order by 4 desc;
 EOF
 
+log "Cleaning flow table..."
 
 REMAINING_ELMTS="${LIMIT}"
 while [ "${REMAINING_ELMTS}" -gt 0 ]; do
@@ -86,7 +90,7 @@ EOF
     log "${OUTPUT}"
 
     if ! [[ "${OUTPUT}" =~ '^DELETE ' ]] ; then
-	log "Output doesn't seems OK..."
+	log "Output doesn't seem OK..."
 	break
     fi
     OUTPUT_NB=$(echo "${OUTPUT}" | cut -d' ' -f 2)
@@ -101,6 +105,73 @@ EOF
     fi
 done
 
+log "Cleaning authentication_session table..."
+
+BATCH_SIZE="${BATCH_SIZE_ORIG}"
+REMAINING_ELMTS="${LIMIT}"
+
+OUTPUT=$(psql "${DSN}" <<EOF
+DO \$\$
+BEGIN
+  DROP TABLE IF EXISTS hydra_cleaner.hydra_childless_auth_session;
+  DROP SCHEMA IF EXISTS hydra_cleaner;
+  CREATE SCHEMA hydra_cleaner;
+  CREATE TABLE hydra_cleaner.hydra_childless_auth_session AS
+  SELECT id
+    FROM hydra_oauth2_authentication_session
+    WHERE NOT EXISTS (
+      SELECT 1
+      FROM hydra_oauth2_flow
+      WHERE hydra_oauth2_authentication_session.id = hydra_oauth2_flow.login_session_id
+    )
+    LIMIT ${REMAINING_ELMTS};
+END \$\$;
+EOF
+)
+
+log "${OUTPUT}"
+
+while [ "${REMAINING_ELMTS}" -gt 0 ]; do
+
+    OUTPUT=$(psql "${DSN}" <<EOF
+WITH cte AS (
+    SELECT id
+    FROM hydra_cleaner.hydra_childless_auth_session
+    LIMIT ${BATCH_SIZE}
+),
+auth_session_deleted as (
+    DELETE
+    FROM hydra_oauth2_authentication_session
+    WHERE hydra_oauth2_authentication_session.id in (select * from cte)
+        AND NOT EXISTS (
+            SELECT 1
+            FROM hydra_oauth2_flow
+            WHERE hydra_oauth2_authentication_session.id = hydra_oauth2_flow.login_session_id
+        )
+    )
+DELETE
+FROM hydra_cleaner.hydra_childless_auth_session
+WHERE hydra_cleaner.hydra_childless_auth_session.id in (select * from cte);
+EOF
+)
+
+    log "${OUTPUT}"
+
+    if ! [[ "${OUTPUT}" =~ '^DELETE ' ]] ; then
+	log "Output doesn't seem OK..."
+	break
+    fi
+    OUTPUT_NB=$(echo "${OUTPUT}" | cut -d' ' -f 2)
+
+    if [ "${OUTPUT_NB}" -lt "${BATCH_SIZE}" ]; then
+	break
+    fi
+
+    REMAINING_ELMTS=$((REMAINING_ELMTS - BATCH_SIZE))
+    if [ "${REMAINING_ELMTS}" -lt "${BATCH_SIZE}" ]; then
+	BATCH_SIZE="${REMAINING_ELMTS}"
+    fi
+done
 
 log "Final estimated size:"
 psql "${DSN}" <<EOF