Shell脚本命令输出与命令行输出不同

Question

OS: RHEL 7.3. I am running below command :

ps uax | grep 'elasticsearch' | grep -v grep | awk '{print $2}'

If I run this command on terminal, I get one PID of Elasticsearch process. However if same command I put inside a shell script like so:

#!/bin/bash

PID=$(ps uax | grep 'elasticsearch' | grep -v grep | awk '{print $2}')
echo $PID

I get several PIDs one below other. What could be going wrong?

Full script which launches ES is as follows:

#!/bin/bash

if [ "$ES_HOME" == "" ]; then
        echo "ES_HOME environment variable does not exists. Please set it to home dir of Elasticsearch and try again"
fi;

PID=$(ps aux | grep "elasticsearch" | grep -vE "start|grep" | awk '{print $2}' | xargs)
if [ "$PID" != "" ]; then
        echo "Elasticsearch is already running with PID: $PID"
        echo ""
        exit;
fi;

echo "Starting Elasticsearch"
sh $ES_HOME/bin/elasticsearch.sh > /dev/null &

Answer 1

In your script you are running it in bash shell #!/bin/bash , just check what is your current shell when you execute manually. If different change the shell accordingly. Otherwise I do not see any difference. Also check what user ID you are using while running the script.

Answer 2

if pgrep is unavailable on your machine, you can use ps without grep to get only the pid of a process using its name .

For example,

 #!/bin/bash

_pid=$(ps -C NetworkManager -o pid=)
echo "$_pid"

NetworkManager is the name of the command that one will see in the process table, when using ps . You can replace that with elasticsearch .