Related
I know it is not recommended, but is it at all possible to pass the user's password to scp?
I'd like to copy a file via scp as part of a batch job and the receiving server does, of course, need a password and, no, I cannot easily change that to key-based authentication.
Use sshpass:
sshpass -p "password" scp -r user#example.com:/some/remote/path /some/local/path
or so the password does not show in the bash history
sshpass -f "/path/to/passwordfile" scp -r user#example.com:/some/remote/path /some/local/path
The above copies contents of path from the remote host to your local.
Install :
ubuntu/debian
apt install sshpass
centos/fedora
yum install sshpass
mac w/ macports
port install sshpass
mac w/ brew
brew install https://raw.githubusercontent.com/kadwanev/bigboybrew/master/Library/Formula/sshpass.rb
just generate a ssh key like:
ssh-keygen -t rsa -C "your_email#youremail.com"
copy the content of ~/.ssh/id_rsa.pub
and lastly add it to the remote machines ~/.ssh/authorized_keys
make sure remote machine have the permissions 0700 for ~./ssh folder and 0600 for ~/.ssh/authorized_keys
If you are connecting to the server from Windows, the Putty version of scp ("pscp") lets you pass the password with the -pw parameter.
This is mentioned in the documentation here.
curl can be used as a alternative to scp to copy a file and it supports a password on the commandline.
curl --insecure --user username:password -T /path/to/sourcefile sftp://desthost/path/
You can script it with a tool like expect (there are handy bindings too, like Pexpect for Python).
You can use the 'expect' script on unix/terminal
For example create 'test.exp' :
#!/usr/bin/expect
spawn scp /usr/bin/file.txt root#<ServerLocation>:/home
set pass "Your_Password"
expect {
password: {send "$pass\r"; exp_continue}
}
run the script
expect test.exp
I hope that helps.
You may use ssh-copy-id to add ssh key:
$which ssh-copy-id #check whether it exists
If exists:
ssh-copy-id "user#remote-system"
Here is an example of how you do it with expect tool:
sub copyover {
$scp = Expect->spawn("/usr/bin/scp ${srcpath}/$file $who:${destpath}/$file");
$scp->expect(30,"ssword: ") || die "Never got password prompt from $dest:$!\n";
print $scp 'password' . "\n";
$scp->expect(30,"-re",'$\s') || die "Never got prompt from parent system:$!\n";
$scp->soft_close();
return;
}
Nobody mentioned it, but Putty scp (pscp) has a -pw option for password.
Documentation can be found here: https://the.earth.li/~sgtatham/putty/0.67/htmldoc/Chapter5.html#pscp
Once you set up ssh-keygen as explained above, you can do
scp -i ~/.ssh/id_rsa /local/path/to/file remote#ip.com:/path/in/remote/server/
If you want to lessen typing each time, you can modify your .bash_profile file and put
alias remote_scp='scp -i ~/.ssh/id_rsa /local/path/to/file remote#ip.com:/path/in/remote/server/
Then from your terminal do source ~/.bash_profile. Afterwards if you type remote_scp in your terminal it should run the scp command without password.
Here's a poor man's Linux/Python/Expect-like example based on this blog post: Upgrading simple shells to fully interactive
TTYs. I needed this for old machines where I can't install Expect or add modules to Python.
Code:
(
echo 'scp jmudd#mysite.com:./install.sh .'
sleep 5
echo 'scp-passwd'
sleep 5
echo 'exit'
) |
python -c 'import pty; pty.spawn("/usr/bin/bash")'
Output:
scp jmudd#mysite.com:install.sh .
bash-4.2$ scp jmudd#mysite.com:install.sh .
Password:
install.sh 100% 15KB 236.2KB/s 00:00
bash-4.2$ exit
exit
Make sure password authentication is enabled on the target server. If it runs Ubuntu, then open /etc/ssh/sshd_config on the server, find lines PasswordAuthentication=no and comment all them out (put # at the start of the line), save the file and run sudo systemctl restart ssh to apply the configuration. If there is no such line then you're done.
Add -o PreferredAuthentications="password" to your scp command, e.g.:
scp -o PreferredAuthentications="password" /path/to/file user#server:/destination/directory
make sure you have "expect" tool before, if not, do it
# apt-get install expect
create the a script file with following content. (# vi /root/scriptfile)
spawn scp /path_from/file_name user_name_here#to_host_name:/path_to
expect "password:"
send put_password_here\n;
interact
execute the script file with "expect" tool
# expect /root/scriptfile
copy files from one server to other server ( on scripts)
Install putty on ubuntu or other Linux machines. putty comes with pscp. we can copy files with pscp.
apt-get update
apt-get install putty
echo n | pscp -pw "Password#1234" -r user_name#source_server_IP:/copy_file_path/files /path_to_copy/files
For more options see pscp help.
Using SCP non interactively from Windows:
Install the community Edition of netcmdlets
Import Module
Use Send-PowerShellServerFile -AuthMode password -User MyUser -Password not-secure -Server YourServer -LocalFile C:\downloads\test.txt -RemoteFile C:\temp\test.txt for sending File with non-interactive password
In case if you observe a strict host key check error then use -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null options.
The complete example is as follows
sshpass -p "password" scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null root#domain-name.com:/tmp/from/psoutput /tmp/to/psoutput
You can use below steps. This works for me!
Step1-
create a normal file suppose "fileWithScpPassword" which contains the ssh password for the destination server.
Step2- use sshpaas -f followed by password file name and then normal scp command.
sshpass -f "fileWithScpPassword" scp /filePathToUpload user#ip:/destinationPath/
One easy way I do this:
Use the same scp cmd as you use with ssh keys i.e
scp -C -i <path_to opens sshkey> <'local file_path'> user#<ip_address_VM>: <'remote file_path’>
for transferring file from local to remote
but instead of providing the correct <path_to_opensshkey>, use some garbage path. Due to wrong key path you will be asked for password instead and you can simply pass the password now to get the work done!
An alternative would be add the public half of the user's key to the authorized-keys file on the target system. On the system you are initiating the transfer from, you can run an ssh-agent daemon and add the private half of the key to the agent. The batch job can then be configured to use the agent to get the private key, rather than prompting for the key's password.
This should be do-able on either a UNIX/Linux system or on Windows platform using pageant and pscp.
All the solutions mentioned above can work only if you the app installed or you should have the admin rights to install except or sshpass.
I found this very useful link to simply start the scp in Background.
$ nohup scp file_to_copy user#server:/path/to/copy/the/file > nohup.out 2>&1
https://charmyin.github.io/scp/2014/10/07/run-scp-in-background/
I found this really helpful answer here.
rsync -r -v --progress -e ssh user#remote-system:/address/to/remote/file /home/user/
Not only you can pass there the password, but also it will show the progress bar when copying. Really awesome.
I know it is not recommended, but is it at all possible to pass the user's password to scp?
I'd like to copy a file via scp as part of a batch job and the receiving server does, of course, need a password and, no, I cannot easily change that to key-based authentication.
Use sshpass:
sshpass -p "password" scp -r user#example.com:/some/remote/path /some/local/path
or so the password does not show in the bash history
sshpass -f "/path/to/passwordfile" scp -r user#example.com:/some/remote/path /some/local/path
The above copies contents of path from the remote host to your local.
Install :
ubuntu/debian
apt install sshpass
centos/fedora
yum install sshpass
mac w/ macports
port install sshpass
mac w/ brew
brew install https://raw.githubusercontent.com/kadwanev/bigboybrew/master/Library/Formula/sshpass.rb
just generate a ssh key like:
ssh-keygen -t rsa -C "your_email#youremail.com"
copy the content of ~/.ssh/id_rsa.pub
and lastly add it to the remote machines ~/.ssh/authorized_keys
make sure remote machine have the permissions 0700 for ~./ssh folder and 0600 for ~/.ssh/authorized_keys
If you are connecting to the server from Windows, the Putty version of scp ("pscp") lets you pass the password with the -pw parameter.
This is mentioned in the documentation here.
curl can be used as a alternative to scp to copy a file and it supports a password on the commandline.
curl --insecure --user username:password -T /path/to/sourcefile sftp://desthost/path/
You can script it with a tool like expect (there are handy bindings too, like Pexpect for Python).
You can use the 'expect' script on unix/terminal
For example create 'test.exp' :
#!/usr/bin/expect
spawn scp /usr/bin/file.txt root#<ServerLocation>:/home
set pass "Your_Password"
expect {
password: {send "$pass\r"; exp_continue}
}
run the script
expect test.exp
I hope that helps.
You may use ssh-copy-id to add ssh key:
$which ssh-copy-id #check whether it exists
If exists:
ssh-copy-id "user#remote-system"
Here is an example of how you do it with expect tool:
sub copyover {
$scp = Expect->spawn("/usr/bin/scp ${srcpath}/$file $who:${destpath}/$file");
$scp->expect(30,"ssword: ") || die "Never got password prompt from $dest:$!\n";
print $scp 'password' . "\n";
$scp->expect(30,"-re",'$\s') || die "Never got prompt from parent system:$!\n";
$scp->soft_close();
return;
}
Nobody mentioned it, but Putty scp (pscp) has a -pw option for password.
Documentation can be found here: https://the.earth.li/~sgtatham/putty/0.67/htmldoc/Chapter5.html#pscp
Once you set up ssh-keygen as explained above, you can do
scp -i ~/.ssh/id_rsa /local/path/to/file remote#ip.com:/path/in/remote/server/
If you want to lessen typing each time, you can modify your .bash_profile file and put
alias remote_scp='scp -i ~/.ssh/id_rsa /local/path/to/file remote#ip.com:/path/in/remote/server/
Then from your terminal do source ~/.bash_profile. Afterwards if you type remote_scp in your terminal it should run the scp command without password.
Here's a poor man's Linux/Python/Expect-like example based on this blog post: Upgrading simple shells to fully interactive
TTYs. I needed this for old machines where I can't install Expect or add modules to Python.
Code:
(
echo 'scp jmudd#mysite.com:./install.sh .'
sleep 5
echo 'scp-passwd'
sleep 5
echo 'exit'
) |
python -c 'import pty; pty.spawn("/usr/bin/bash")'
Output:
scp jmudd#mysite.com:install.sh .
bash-4.2$ scp jmudd#mysite.com:install.sh .
Password:
install.sh 100% 15KB 236.2KB/s 00:00
bash-4.2$ exit
exit
Make sure password authentication is enabled on the target server. If it runs Ubuntu, then open /etc/ssh/sshd_config on the server, find lines PasswordAuthentication=no and comment all them out (put # at the start of the line), save the file and run sudo systemctl restart ssh to apply the configuration. If there is no such line then you're done.
Add -o PreferredAuthentications="password" to your scp command, e.g.:
scp -o PreferredAuthentications="password" /path/to/file user#server:/destination/directory
make sure you have "expect" tool before, if not, do it
# apt-get install expect
create the a script file with following content. (# vi /root/scriptfile)
spawn scp /path_from/file_name user_name_here#to_host_name:/path_to
expect "password:"
send put_password_here\n;
interact
execute the script file with "expect" tool
# expect /root/scriptfile
copy files from one server to other server ( on scripts)
Install putty on ubuntu or other Linux machines. putty comes with pscp. we can copy files with pscp.
apt-get update
apt-get install putty
echo n | pscp -pw "Password#1234" -r user_name#source_server_IP:/copy_file_path/files /path_to_copy/files
For more options see pscp help.
Using SCP non interactively from Windows:
Install the community Edition of netcmdlets
Import Module
Use Send-PowerShellServerFile -AuthMode password -User MyUser -Password not-secure -Server YourServer -LocalFile C:\downloads\test.txt -RemoteFile C:\temp\test.txt for sending File with non-interactive password
In case if you observe a strict host key check error then use -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null options.
The complete example is as follows
sshpass -p "password" scp -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null root#domain-name.com:/tmp/from/psoutput /tmp/to/psoutput
You can use below steps. This works for me!
Step1-
create a normal file suppose "fileWithScpPassword" which contains the ssh password for the destination server.
Step2- use sshpaas -f followed by password file name and then normal scp command.
sshpass -f "fileWithScpPassword" scp /filePathToUpload user#ip:/destinationPath/
One easy way I do this:
Use the same scp cmd as you use with ssh keys i.e
scp -C -i <path_to opens sshkey> <'local file_path'> user#<ip_address_VM>: <'remote file_path’>
for transferring file from local to remote
but instead of providing the correct <path_to_opensshkey>, use some garbage path. Due to wrong key path you will be asked for password instead and you can simply pass the password now to get the work done!
An alternative would be add the public half of the user's key to the authorized-keys file on the target system. On the system you are initiating the transfer from, you can run an ssh-agent daemon and add the private half of the key to the agent. The batch job can then be configured to use the agent to get the private key, rather than prompting for the key's password.
This should be do-able on either a UNIX/Linux system or on Windows platform using pageant and pscp.
All the solutions mentioned above can work only if you the app installed or you should have the admin rights to install except or sshpass.
I found this very useful link to simply start the scp in Background.
$ nohup scp file_to_copy user#server:/path/to/copy/the/file > nohup.out 2>&1
https://charmyin.github.io/scp/2014/10/07/run-scp-in-background/
I found this really helpful answer here.
rsync -r -v --progress -e ssh user#remote-system:/address/to/remote/file /home/user/
Not only you can pass there the password, but also it will show the progress bar when copying. Really awesome.
I have a text file, like this:
cat hed.txt
10.21.23.12
10.23.12.12
I can ssh to each ip without without prompting for the key verification.
I want to run a command on each of these IPs, so I was using pdsh. I tried multiple options, but I am getting following error:
pdsh -w ^hed uptime
00f12e86-cfcc-4239-9dfc-006b65a319c3: ssh: Could not resolve hostname 00f12e86-cfcc-4239-9dfc-006b65a319c3: nodename nor servname provided, or not known
pdsh#saurabh: 00f12e86-cfcc-4239-9dfc-006b65a319c3: ssh exited with exit code 255
I mentioned here, I tried following as well, but this also gave same error.
PDSH_SSH_ARGS_APPEND="-o StrictHostKeyChecking=no" pdsh -R ssh -w ^hed uptime
Also tried comment from here, but no help.
PDSH_SSH_ARGS_APPEND="-o StrictHostKeyChecking=no" pdsh -R ssh ^hed uptime
pdsh#saurabh: no remote hosts specified
I am able to do csshx on these via: csshX --host hed.txt, which works but pdsh will suit more for my work which is not working.
Ahh, This worked like this:
pdsh -w '^hed.txt' uptime
For my colleagues it is working without quotes as well with same version of pdsh, which is weird.
I have docker-engine installed on a linux VM on my company data center. I installed docker-machine on my windows. I want to manage this docker-engine through my windows machine. I want to add this host and for that I executed the following command:
docker-machine create -d generic --generic-ip-address 10.51.227.5 --generic-ssh-port 22 --generic-ssh-user root compute
But I am getting the following error after waiting for couple of minutes
Running pre-create checks...
Creating machine...
(compute) No SSH key specified. Connecting to this machine now and in the future will require the ssh agent to contain the appropriate key.
Waiting for machine to be running, this may take a few minutes...
Detecting operating system of created instance...
Waiting for SSH to be available...
Error creating machine: Error detecting OS: Too many retries waiting for SSH to be available. Last error: Maximum number of retries (60) exceeded
I don't know what I am doing wrong.
Try using native SSH
Unfortunately, this can happen for multiple reasons, but since you say you're running on Windows, I have a guess. Try using the native SSH client - that is, the Go library that ships with Docker, rather than ssh.exe. To do so, pass the --native-ssh argument to docker-machine:
docker-machine --native-ssh create --driver generic --generic-ip-address example.com ExampleMachine
Examining debug output
You can examine various debugging messages by passing the --debug argument. When I did that, and did not specify the --native-ssh argument, I got lots of error messages:
> docker-machine create --driver generic --generic-ip-address example.com ExampleMachine
<...snip...>
Waiting for SSH to be available...
Getting to WaitForSSH function...
(apsylania) Calling .GetSSHHostname
(apsylania) Calling .GetSSHPort
(apsylania) Calling .GetSSHKeyPath
(apsylania) Calling .GetSSHKeyPath
(apsylania) Calling .GetSSHUsername
Using SSH client type: external
&{[-F /dev/null -o PasswordAuthentication=no -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -o LogLevel=quiet -o ConnectionAttempts=3 -o ConnectTimeout=10 -o ControlMaster=no -o ControlPath=none root#example.com -o IdentitiesOnly=yes -i C:\Users\MyAccount\.docker\machine\machines\ExampleMachine\id_rsa -p 22] C:\Program Files\OpenSSH-Win64\ssh.exe <nil>}
About to run SSH command:
exit 0
SSH cmd err, output: exit status 255:
Error getting ssh command 'exit 0' : ssh command error:
command : exit 0
err : exit status 255
output :
Getting to WaitForSSH function...
<...snip...>
My guess, after trying to pass --generic-ssh-key with backslashes on Windows and having it produce an error, was that Docker (still) isn't quite as polished on Windows as it could be, and I wondered if there was something buggy about the way it calls out to an external SSH client. Since using the native Go SSH client worked for me, I think that my guess was correct, at least for my case.
I just trying ouy a playbook that a colleague has set up that I needed to modify. The first problem I get running on my mac was
ERROR: Unable to find an inventory file, specify one with -i ?
This was easily solved by adding -i verif to the command. But then the following error occured.
loadgen-verif-app1.internal.machines | FAILED => SSH encountered an unknown error. The output was:
OpenSSH_6.2p2, OSSLShim 0.9.8r 8 Dec 2011
debug1: Reading configuration data /Users/andreas.joelsson/.ssh/config
debug1: Reading configuration data /etc/ssh_config
debug1: /etc/ssh_config line 20: Applying options for *
debug1: auto-mux: Trying existing master
ControlPath too long
This is true for all 8 machines (loadgen-verif-app[1-8].internal.machines)
After some debugging that the file could be too long, I tried the following command with the same result:
ansible nukes -m ping -i verif -vvvv
then I thought it was issues with ssh but executing the command through ssh works:
ssh loadgen-verif-app1.internal.machines ping loadgen-verif-app2.internal.machines
And now I am stumped because the ping command works on some of the machines not in the range listed above, the thing is that they are shorter than the loagenXXX.machines path if that makes it an issue. But then the ssh command shouldn't work I guess.
I have some ssh config settings set up for the targets as well, but that is no different that the ones that did work with the ping command.
Host loadgen1
HostName loadgen-verif-app1.internal.machines
Now I am stumped as it works for the colleague on a mac as well. So not sure if there is some setting i'm missing or similar. He doesn't need to provide the -i verif either that can also be a reason why it doesn't work.
edit 2014-12-17:
Have tried modifying the ansible setting control_path according to http://docs.ansible.com/intro_configuration.html#control-path
We are running the same version of ansible
We are running the same version of OpenSSH.
We have the same ssh configs as far as we can tell.
Have been looking for Host* that I found in /etc/ssh_config and removed without progress according to f.e. https://help.openshift.com/hc/en-us/articles/202186044-Unable-to-git-clone-an-application-when-SSH-session-sharing-is-in-use-ControlPath-too-long-
edit 2015-01-08:
SE-C02N76PGG5RP:verif_provisioning andreas.joelsson$ ansible loadgen-verif-app1.internal.machines -m ping -i verif -vvvv
<loadgen-verif-app1.internal.machines> ESTABLISH CONNECTION FOR USER: andreas.joelsson
<loadgen-verif-app1.internal.machines> REMOTE_MODULE ping
<loadgen-verif-app1.internal.machines> EXEC ssh -C -tt -vvv -o ControlMaster=auto -o ControlPersist=60s -o ControlPath="/Users/andreas.joelsson/.ansible/cp/ansible-ssh-%h-%p-%r" -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o ConnectTimeout=10 loadgen-verif-app1.internal.machines /bin/sh -c 'mkdir -p $HOME/.ansible/tmp/ansible-tmp-1420723708.99-33622628424665 && chmod a+rx $HOME/.ansible/tmp/ansible-tmp-1420723708.99-33622628424665 && echo $HOME/.ansible/tmp/ansible-tmp-1420723708.99-33622628424665'
loadgen-verif-app1.internal.machines | FAILED => SSH encountered an unknown error. The output was:
OpenSSH_6.2p2, OSSLShim 0.9.8r 8 Dec 2011
debug1: Reading configuration data /Users/andreas.joelsson/.ssh/config
debug1: Reading configuration data /etc/ssh_config
debug1: /etc/ssh_config line 20: Applying options for *
debug1: auto-mux: Trying existing master
ControlPath too long
edit 2015-02-12:
SE-C02N76PGG5RP:verif_provisioning andreas.joelsson$ ansible nukes -m ping -i verif
loadgen-verif-app4.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app5.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app3.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app1.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app2.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app8.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app6.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
loadgen-verif-app7.internal.machines | FAILED => SSH encountered an unknown error during the connection. We recommend you re-run the command using -vvvv, which will enable SSH debugging output to help diagnose the issue
And with the working one:
SE-C02N76PGG5RP:verif_provisioning andreas.joelsson$ ansible duke -m ping -i verif
steve-verif-app1.internal.machines | success >> {
"changed": false,
"ping": "pong"
}
Solution to this error is mentioned at ansible documentation, please refer to this link.
I was getting this error, when I tried to connect to EC2 instances but after modifying the below mentioned configuration, it solved my problem.
I am supposing that you have installed the ansible on Mac using pip. So, please do these steps:
create the /etc/ansible directory
sudo mkdir /etc/ansible
change the permission of it
sudo chown $(whoami):staff /etc/ansible
download the ansible.cfg file from here and place inside the /etc/ansible directory
edit/uncomment the following line
[ssh_connection]
control_path = %(directory)s/%%h-%%r
edit the ~/.ssh/config file:
Host *
GSSAPIAuthentication no
EXTRA STEP:
brew install https://raw.github.com/eugeneoden/homebrew/eca9de1/Library/Formula/sshpass.rb
Ok the way I got it working was that I did the changes in ansible.cfg and the extra steps but it did not work. The only way I found is to export ANSIBLE_SSH_CONTROL_PATH:
This is coz, I think it always is picking the default path. Even after the change in ansible.cfg
1.9.4 git:(master) pwd
/usr/local/Cellar/ansible/1.9.4
➜ 1.9.4 git:(master) ag ANSIBLE_SSH_CONTROL
libexec/lib/python2.7/site-packages/ansible/constants.py
187:ANSIBLE_SSH_CONTROL_PATH = get_config(p, 'ssh_connection', 'control_path', 'ANSIBLE_SSH_CONTROL_PATH', "%(directory)s/ansible-ssh-%%h-%%p-%%r")
Output without exporting ANSIBLE_SSH_CONTOL_PATH:
ControlPath="/Users/vinitkhandagle/.ansible/cp/ansible-ssh-%h-%p-%r"
Exported the variable as:
export ANSIBLE_SSH_CONTROL_PATH='%(directory)s/%%h-%%r'
Control path changes accordingly:
ControlPath="/Users/vinitkhandagle/.ansible/cp/%h-%r"
Adding a note to #techraf above comment is that I encountered the same issue when using ansible in combination with molecule in a virtual environment. So even though ansible.cfg has been read. ANSIBLE_SSH_CONTROL_PATH will be overwritten. So in that case a workaround was to set it as environment variable outside the configuration file itself. As this is then shorter than the default path.
export ANSIBLE_SSH_CONTROL_PATH='%(directory)s/tmp'