From d21e3628618d33c0557fae00989f85c4fce9c1a3 Mon Sep 17 00:00:00 2001
From: Cheng Pan <chengpan@apache.org>
Date: Fri, 15 Nov 2024 19:17:36 +0800
Subject: [PATCH] [ZEPPELIN-6150] `zeppelin.sh` shall respect HADOOP_HOME to
 find `hadoop` command

### What is this PR for?

Obviously, `*_HOME` should be respected if they are present, otherwise fallback to find the command in `PATH`.


### What type of PR is it?

Improvement

### What is the Jira issue?

ZEPPELIN-6150

### How should this be tested?

Manually tested on a dev env, there are multiple Hadoop clients installed in one machine, and `hadoop` is not available in `PATH`, now I can set `HADOOP_HOME` to allow Zeppelin start.

### Screenshots (if appropriate)

### Questions:
* Does the license files need to update? no
* Is there breaking changes for older versions? might be
* Does this needs documentation? no


Closes #4899 from pan3793/ZEPPELIN-6150.

Signed-off-by: Cheng Pan <chengpan@apache.org>
---
 bin/zeppelin.sh | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/bin/zeppelin.sh b/bin/zeppelin.sh
index efd3aae8b8a..9c1f7a9b41e 100755
--- a/bin/zeppelin.sh
+++ b/bin/zeppelin.sh
@@ -121,7 +121,9 @@ if [[ "${USE_HADOOP}" != "false"  ]]; then
     echo "Please specify HADOOP_CONF_DIR if USE_HADOOP is true"
   else
     ZEPPELIN_CLASSPATH+=":${HADOOP_CONF_DIR}"
-    if ! [ -x "$(command -v hadoop)" ]; then
+    if [ -n "${HADOOP_HOME}" ]; then
+      ZEPPELIN_CLASSPATH+=":`${HADOOP_HOME}/bin/hadoop classpath`"
+    elif ! [ -x "$(command -v hadoop)" ]; then
       echo 'hadoop command is not in PATH when HADOOP_CONF_DIR is specified.'
     else
       ZEPPELIN_CLASSPATH+=":`hadoop classpath`"