From 1c84daec6c2e717c3f2f7f9e6497124272e3195f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Gaspard=20F=C3=A9rey?= <gaspard.ferey@gmail.com>
Date: Thu, 16 Jan 2025 17:15:30 +0100
Subject: [PATCH] [Spark] Update to Spark 3.5.4 (#237)

---
 .github/workflows/main-workflow.yml  | 10 +++++-----
 scripts/install-spark-hadoop-hive.sh |  4 ++--
 spark/Dockerfile                     |  2 +-
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/.github/workflows/main-workflow.yml b/.github/workflows/main-workflow.yml
index 751fb21..460eaf1 100644
--- a/.github/workflows/main-workflow.yml
+++ b/.github/workflows/main-workflow.yml
@@ -103,7 +103,7 @@ jobs:
       base_image: python-minimal
       python_version_1: 3.12.7
       python_version_2: 3.11.10
-      spark_version: 3.5.3
+      spark_version: 3.5.4
       build_gpu: false
     secrets: inherit
   jupyter-pyspark:
@@ -115,7 +115,7 @@ jobs:
       base_image: pyspark
       python_version_1: 3.12.7
       python_version_2: 3.11.10
-      spark_version: 3.5.3
+      spark_version: 3.5.4
       build_gpu: false
     secrets: inherit
   vscode-pyspark:
@@ -127,7 +127,7 @@ jobs:
       base_image: pyspark
       python_version_1: 3.12.7
       python_version_2: 3.11.10
-      spark_version: 3.5.3
+      spark_version: 3.5.4
       build_gpu: false
     secrets: inherit
   jupyter-pytorch:
@@ -189,7 +189,7 @@ jobs:
       base_image: r-minimal
       r_version_1: 4.4.2
       r_version_2: 4.3.3
-      spark_version: 3.5.3
+      spark_version: 3.5.4
       build_gpu: false
     secrets: inherit
   rstudio-sparkr:
@@ -201,7 +201,7 @@ jobs:
       base_image: sparkr
       r_version_1: 4.4.2
       r_version_2: 4.3.3
-      spark_version: 3.5.3
+      spark_version: 3.5.4
       build_gpu: false
     secrets: inherit
   jupyter-r:
diff --git a/scripts/install-spark-hadoop-hive.sh b/scripts/install-spark-hadoop-hive.sh
index 7771b6f..33b7910 100644
--- a/scripts/install-spark-hadoop-hive.sh
+++ b/scripts/install-spark-hadoop-hive.sh
@@ -1,7 +1,7 @@
 #!/bin/bash
 set -e
 
-HADOOP_VERSION="3.4.0"
+HADOOP_VERSION="3.4.1"
 HIVE_VERSION="2.3.9"
 HIVE_LISTENER_VERSION="0.0.3"
 
@@ -48,4 +48,4 @@ mv jline-2.14.6.jar ${HIVE_HOME}/lib/
 rm ${HIVE_HOME}/lib/jline-2.12.jar
 
 # Fix multiple bindings
-rm ${HADOOP_HOME}/share/hadoop/tools/lib/bundle-2.23.19.jar
+rm ${HADOOP_HOME}/share/hadoop/tools/lib/bundle-2.24.6.jar
diff --git a/spark/Dockerfile b/spark/Dockerfile
index 1b47af0..d56e481 100644
--- a/spark/Dockerfile
+++ b/spark/Dockerfile
@@ -5,7 +5,7 @@ LABEL maintainer="InseeFrLab <innovation@insee.fr>"
 
 ARG BASE_IMAGE
 
-ARG SPARK_VERSION="3.5.3"
+ARG SPARK_VERSION="3.5.4"
 ENV SPARK_VERSION=${SPARK_VERSION}
 ENV JAVA_VERSION="17"