SSH pseudo terminal zombied after ansible playbook exit

Archives Ansible Developer
,
1

I am running ansible-playbook inside docker container on kubernetes. It does copy task and executes a script. Uses private keys. I am invoking it using a java Process. Looks like the ssh pseudo terminals it might be opening stay behind once the playbook is finished executing. Any new playbook invocation adds on to the number of ssh processes. It sort of what this thread mentions i think, only that I don’t see -tt option being appended. I tried turning pty false(which is needed for sftp anyways i think) in ansible.cfg as well. Nothing improves. Once ansible-playbook process dies i will see bunch of defunct [ssh] around.
Here are couple lines from the “ps” output

root 1627 1 0 00:00 ? 00:00:00 ssh: /root/.ansible/cp/c360b1d939 [mux]
root 1629 1 0 00:00 ? 00:00:00 [ssh]
root 1630 1 0 00:00 ? 00:00:00 ssh: /root/.ansible/cp/e8f6c354e2 [mux]
root 1632 1 0 00:00 ? 00:00:00 [ssh]
root 1633 1 0 00:00 ? 00:00:00 ssh: /root/.ansible/cp/edb897f2b6 [mux]
root 1643 1641 0 00:00 ? 00:00:00 ssh -C -o ControlMaster=auto -o ControlPersist=60s -o StrictHostKeyChecking=no -o IdentityFile=“/opt/ansible/id_rsa” -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o User= -o ConnectTimeout=5 -o ControlPath=/root/.ansible/cp/b1739bcf8c /bin/sh -c ‘/usr/bin/python && sleep 0’
root 1645 1644 0 00:00 ? 00:00:00 ssh -C -o ControlMaster=auto -o ControlPersist=60s -o StrictHostKeyChecking=no -o IdentityFile=“/opt/ansible/id_rsa” -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o User= -o ConnectTimeout=5 -o ControlPath=/root/.ansible/cp/d54b353787 /bin/sh -c ‘/usr/bin/python && sleep 0’

Top output

1 root 20 0 14.980g 1.487g 15156 S 1.3 0.9 4:17.71 java
9377 root 20 0 38376 3440 1248 R 1.0 0.0 0:00.20 top
449 root 20 0 0 0 0 Z 0.0 0.0 0:00.00 ssh
450 root 20 0 0 0 0 Z 0.0 0.0 0:00.04 ssh
453 root 20 0 0 0 0 Z 0.0 0.0 0:00.00 ssh
454 root 20 0 0 0 0 Z 0.0 0.0 0:00.04 ssh
456 root 20 0 0 0 0 Z 0.0 0.0 0:00.00 ssh

Code to invoke playbook:

String cmdPurge = { “ansible-playbook”, config.getPurgePlaybookPath(), “-i”, config.getInventoryLoc(),
“–extra-vars”,
“{"urls_file":” + config.getUrlPath() + “,"type":"” + config.getType() + “","purge_file":"” + config.getPurgeFile() + “"}” ,“–private-key”,config.getPrivateKeyFile()};
Process pPurge = Runtime.getRuntime().exec(cmdPurge, null);
BufferedReader in = new BufferedReader(new InputStreamReader(pPurge.getInputStream()));
while ((line = in.readLine()) != null) {
//process the console output
}
in.close();
int exitcode = pPurge.waitFor();

Ansible.cfg

inventory = hosts
remote_user =
host_key_checking = False
timeout = 5
pty=False
pipelining = True

Base Image: Ubuntu:Xenial

Ansible version : ansible 2.7.2

using JDK8

I faced same issues using Alpine jdk image and then I switched to ubuntu.

2

Adding the playbook code

- name: "copying URL file"
  copy: src="{{urls_file}}" dest="~/{{urls_file}}"

- name: "copying purge script"
  copy: src="{{purge_file}}" dest="~/{{purge_file}}" mode=0777

- name: "execute purge script"
  shell: ./"{{purge_file}}" "{{urls_file}}"
  register: machine

- debug: msg="{{machine.stdout_lines}}"

- name: "Clean up copied file"
  file:
    path: "{{item}}"
    state: absent
  with_items:
    - "{{urls_file}}"
    - "{{purge_file}}"