From 00755e30495b5f32ca9b5eaee74d1ee60c8ccd69 Mon Sep 17 00:00:00 2001
From: Fabian Haas <29468630+hfxbse@users.noreply.github.com>
Date: Tue, 19 Nov 2024 18:10:24 +0100
Subject: [PATCH] Connect to spark independent of the host name

---
 airflow.Dockerfile | 7 +++++--
 docker-compose.yml | 3 +++
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/airflow.Dockerfile b/airflow.Dockerfile
index 85a3506..5d51c83 100644
--- a/airflow.Dockerfile
+++ b/airflow.Dockerfile
@@ -11,9 +11,12 @@ RUN wget https://jdbc.postgresql.org/download/postgresql-42.7.4.jar -P  /home/ai
 RUN mv /tmp/upstream/exercises/winter_semester_2024-2025/05_airflow/plugins /home/airflow/airflow
 RUN mv /tmp/upstream/exercises/winter_semester_2024-2025/05_airflow/dags /home/airflow/airflow
 RUN mv /tmp/upstream/exercises/winter_semester_2024-2025/05_airflow/python /home/airflow/airflow
+
+RUN sed -i '34 i sed -i "s/SPARK_MASTER_IP=hadoop/SPARK_MASTER_IP=\$HADOOP_HOST/" /home/airflow/spark/conf/spark-env.sh' /startup.sh
+RUN sed -i '34 i sed -i "s/SPARK_LOCAL_IP=\\"airflow\\"/SPARK_LOCAL_IP=\$SPARK_HOST/" /home/airflow/spark/conf/spark-env.sh' /startup.sh
 # Pyarrow fails to interperet placeholder, subsitute instead
-RUN sed -i "34 i sed -i 's/hadoop:/$HADOOP_HOST:/g' /home/airflow/hadoop/etc/hadoop/core-site.xml" /startup.sh
-RUN sed -i "34 i sed -i 's/hadoop:/$HADOOP_HOST:/g' /home/airflow/hadoop/etc/hadoop/yarn-site.xml" /startup.sh
+RUN sed -i '34 i sed -i "s/hadoop:/\$HADOOP_HOST:/g" /home/airflow/hadoop/etc/hadoop/core-site.xml' /startup.sh
+RUN sed -i '34 i sed -i "s/hadoop:/\$HADOOP_HOST:/g" /home/airflow/hadoop/etc/hadoop/yarn-site.xml' /startup.sh
 # Setting AIRFLOW__WEBSERVER__BASE_URL did not get applied for unknown reasons, update the config file instead
 RUN sed -i "34 i sed -i 's#base_url = http://localhost:8080#base_url = http://localhost:8080/airflow#' /home/airflow/airflow/airflow.cfg" /startup.sh
 
diff --git a/docker-compose.yml b/docker-compose.yml
index 64064c5..e764702 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -42,3 +42,6 @@ services:
         condition: service_started
       hadoop:
         condition: service_healthy
+    environment:
+      - HADOOP_HOST=hadoop
+      - SPARK_HOST=airflow    # Define the address under which spark is reachable externally