如何使用参数通过 ssh 执行远程命令？

Question

In my .bashrc I define a function which I can use on the command line later:

function mycommand() {
    ssh user@123.456.789.0 cd testdir;./test.sh "$1"
}

When using this command, just the cd command is executed on the remote host; the test.sh command is executed on the local host. This is because the semicolon separates two different commands: the ssh command and the test.sh command.

I tried defining the function as follows (note the single quotes):

function mycommand() {
    ssh user@123.456.789.0 'cd testdir;./test.sh "$1"'
}

I tried to keep the cd command and the test.sh command together, but the argument $1 is not resolved, independent of what I give to the function. It is always tried to execute a command

./test.sh $1

on the remote host.

How do I properly define mycommand , so the script test.sh is executed on the remote host after changing into the directory testdir , with the ability to pass on the argument given to mycommand to test.sh ?

Answer 1

Do it this way instead:

function mycommand {
    ssh user@123.456.789.0 "cd testdir;./test.sh \"$1\""
}

You still have to pass the whole command as a single string, yet in that single string you need to have $1 expanded before it is sent to ssh so you need to use "" for it.

Update

Another proper way to do this actually is to use printf %q to properly quote the argument. This would make the argument safe to parse even if it has spaces, single quotes, double quotes, or any other character that may have a special meaning to the shell:

function mycommand {
    printf -v __ %q "$1"
    ssh user@123.456.789.0 "cd testdir;./test.sh $__"
}

• When declaring a function with function , () is not necessary.
• Don't comment back about it just because you're a POSIXist.

Answer 2

Reviving an old thread, but this pretty clean approach was not listed.

function mycommand() {
    ssh user@123.456.789.0 <<+
    cd testdir;./test.sh "$1"
+
}

Answer 3

This is an example that works on the AWS Cloud. The scenario is that some machine that booted from autoscaling needs to perform some action on another server, passing the newly spawned instance DNS via SSH

# Get the public DNS of the current machine (AWS specific)
MY_DNS=`curl -s http://169.254.169.254/latest/meta-data/public-hostname`


ssh \
    -o StrictHostKeyChecking=no \
    -i ~/.ssh/id_rsa \
    user@remotehost.example.com \
<< EOF
cd ~/
echo "Hey I was just SSHed by ${MY_DNS}"
run_other_commands
# Newline is important before final EOF!

EOF

Answer 4

我正在使用以下命令在本地计算机上执行远程命令：

ssh -i ~/.ssh/$GIT_PRIVKEY user@$IP "bash -s" < localpath/script.sh $arg1 $arg2

Answer 5

A little trick for me, using the "bash -s" they said they allow POSITIONAL ARGS but apparently the $0 is already reserved for whatever reason... Then using twice the same args rocks like so:

ssh user@host "bash -s" < ./start_app.sh -e test -e test -f docker-compose.services.yml 

Answer 6

Solution : you want to be able connect to machine remotely using ssh protocol and trigger/run some actions outside.

on ssh use a -t flag, from documentation:

-t Force pseudo-terminal allocation.
This can be used to execute arbitrary screen-based programs on a remote machine, which can be very useful, eg when implementing menu services. Multiple -t options force tty allocation, even if ssh has no local tty. when implementing menu services. Multiple -t options force tty allocation, even if ssh has no local tty.

formula :

ssh -i <key-path> <user>@<remote-machine> -t '<action>'

Example : as administrator I want to be able to connect remotely into ec2 machines and trigger a revert process for a bad deployment on a several machines in a raw, moreover you better implement this action as an automation script that use ips as an arguments and running on different machines in parallel.

ssh -i /home/admin/.ssh/key admin@10.20.30.40 -t 'cd /home/application && make revert'