CODEWITHSUNDEEP
linuxshellssh
Ha Son Hai
Ha Son Hai

Reputation: 176

Execute ssh command inside a script


I try to make a script to automate my experiments.
My purpose is to run a command like this from a shell script which to run another script on the server: 

ssh [email protected] 'sh ControlIperfServer.sh start'

Even though the file ControlIperfServer.sh was already placed in /root directory of the server.com but the output from terminal is: 

bash: sh ControlIperfServer.sh start: command not found

While I was able to run that command when I type it in terminal. It only can not be executed when executing it from a script.

What I mean is:
When I type the command in terminal. It works.

[root@CentOSVanilla ~]# ssh [email protected] 'sh ControlIperfServer.sh start'
Running IPERF SERVER
[root@CentOSVanilla ~]#

When I put the command in the script and execute the script. It does not work. 

[root@CentOSVanilla ~]# ./Runtest.sh
bash: sh ControlIperfServer.sh start: command not found
[root@CentOSVanilla ~]#

I used the verbose -v option to track 

debug1: Next authentication method: publickey
debug1: Trying private key: /root/.ssh/identity
debug1: Offering public key: /root/.ssh/id_rsa
debug1: Server accepts key: pkalg ssh-rsa blen 277
debug1: read PEM private key done: type RSA
debug1: Authentication succeeded (publickey).
debug1: channel 0: new [client-session]
debug1: Requesting [email protected]
debug1: Entering interactive session.
debug1: Sending environment.
debug1: Sending env LANG = en_US.UTF-8
debug1: Sending env LC_CTYPE = UTF-8
debug1: Sending command: 'sh ControlIperfServer.sh start;'
bash: sh ControlIperfServer.sh start;: command not found
debug1: client_input_channel_req: channel 0 rtype exit-status reply 0
debug1: client_input_channel_req: channel 0 rtype [email protected] reply 0
debug1: channel 0: free: client-session, nchannels 1
Transferred: sent 2352, received 2360 bytes, in 0.2 seconds
Bytes per second: sent 10829.0, received 10865.9
debug1: Exit status 127

Thank you very very very much for your help.

I add the code of my Runtest.sh:

#!/bin/sh
# A script to run test automatically, need modify when using
# Usage:
#       ./RunTest.sh <number of test> [threads]
#       [threads]: when there is threads, use -P option of Iperf
#                  Otherwise, run multiple instances of Iperf
#       <number of test>: Defaut is 1
#                         Number of Threads/Processes
#                         Number of time capture traffic
RUN_IPERF_SERVER="sh ControlIperfServer.sh start"
STOP_IPERF_SERVER="sh ControlIperfServer.sh stop"
RUN_TCPDUMP="sh ControlTCPdump.sh start"
STOP_TCPDUMP="sh ControlTCPdump.sh stop"
RUN_CLIENT="iperf -c"
TARGET_SERVER="10.10.10.253"
REMOTE_SERVER="server.com"
DIRECTORY="CaptureTraffic"
TCP_DURATION="60"

if [ "$1" = "" ]; then
    MAX_TEST=1;
else
    MAX_TEST=$1;
fi

# Run server daemon on remote server
CMD="ssh -v root@$REMOTE_SERVER \"$RUN_IPERF_SERVER\""
echo Running $CMD
$CMD # Run iperf server remotely
sleep 5 # Waiting for server to run

CMD="ssh -v $REMOTE_SERVER 'mkdir $DIRECTORY && echo ok'"
DIROK="$($CMD)"
if [ "$DIROK" = ok ]; then
    mkdir $DIRECTORY
else
    echo Cannot create directory\! Please be careful\!
fi

for INCR in $(seq 1 $MAX_TEST)
do
    echo Running Test $INCR

    #Prepare TCP to capture traffic (run on host)
    echo Setting-up TCPdump for capture traffic
    FILENAME="$DIRECTORY/CapturedTraffic$INCR.dmp"
    ssh $REMOTE_SERVER \'$RUN_TCPDUMP $FILENAME\'
    sleep 5

    # Running the client (running locally)
    if [ "$2" = threads ]; then
        echo "Running Iperf Client as threads model"
        $RUN_CLIENT $TARGET_SERVER -t $TCP_DURATION -P $INCR
    else
        echo "Running Iperf Client as processes model"
        sh parallel.sh -j $INCR \"$RUN_CLIENT $TARGET_SERVER -t $TCP_DURATION\"
    fi

    #Stop TCPdump (run on host)
    ssh $REMOTE_SERVER '$STOP_TCPDUMP'    
    sleep 5 #Wait for TCPdump write file to harddrive

done

#Stop Iperf Deamon
ssh $REMOTE_SERVER '$STOP_IPERF_SERVER'
sleep 3

I already placed all the script files on the host named "server.com".

Upvotes: 0

Views: 5841

Answers (3)

chepner
chepner

Reputation: 530902

This

CMD="ssh -v root@$REMOTE_SERVER \"$RUN_IPERF_SERVER\""

is a bad idea. The quotes embeded in CMD are treated as literal quotes; they are not used simply to protect whitespace in RUN_IPERF_SERVER. It's OK to put the name of a command in a variable, but its arguments should be kept separate. If you are using bash (or some other shell that supports arrays), I recommend

CMD=ssh
CMD_OPTIONS=( -v root@REMOTE_SERVER "$RUN_IPERF_SERVER" )
$CMD "${CMD_OPTIONS[@]}"

If you cannot use arrays, you could fall back to using eval:

CMD="ssh -v root@$REMOTE_SERVER \"$RUN_IPERF_SERVER\""
eval "$CMD"

but I would recommend this

CMD=ssh
CMD_OPTIONS="-v root@REMOTE_SERVER"  # Note no whitespace to protect
$CMD $CMD_OPTIONS "$REMOTE_IPERF_SERVER"

Upvotes: 4

edlerd
edlerd

Reputation: 2145

Give /bin/sh a try instead of sh. Your aliases might not be loaded if you call ssh from the script.

Upvotes: -2

Robert Jacobs
Robert Jacobs

Reputation: 3340

Looks like it thinks 'sh ControlIperfServer.sh start' is a command. One file. Not a file with arguments. Try ' escapes like \' to fix.

Also use a either ./ControlIperfServer.sh or /root/ControlIperfServer.sh. A root's shell should never have "." in its path for security reasons, so don't count on it being there.

Upvotes: 0

Related Questions