[ansible-project] Ansible provisioning through bastion SSH host

[a***@edlio.com]

Hi folks,

I'm encountering issues provisioning with Ansible via SSH from outside our
datacenter (well, AWS Virtual Private Cloud).

Our bastion hosts ("jump boxes") are configured to accept SSH keys from
staff members. Boxes behind the bastion hosts accept only the SSH keys
created for the jumpbox user on the bastion hosts. This makes our
administration easier in the event of key compromise staff turnover (only
need to rotate out on the jumpboxes).

The ProxyCommand examples I've seen for Ansible rely on agent forwarding
and netcat. I want Ansible to SSH in to the bastion host, then run a
second SSH command with the private key of the bastion host. Thus agent
forwarding is useless here. I've tried replacing the netcat ProxyCommand
with "ssh -q -t awsjumpbox1 ssh -p %p %h" and I get mysterious failures
that even -vvvv doesn't provide much insight on. For instance:

fatal: [172.31.17.144] => SSH Error: -bash: line 1:
SSH-2.0-OpenSSH_6.6.1_hpn13v11: command not found
It is sometimes useful to re-run the command using -vvvv, which prints SSH
debug output to help diagnose the issue.

Our ansible.cfg looks like:

[defaults]
host_key_checking = False

[ssh_connection]
ssh_args = -o ControlPersist=15m -F ssh.config -q
scp_if_ssh = True
control_path = ~/.ssh/mux-%%r@%%h:%%p

Our ssh.config (simplified) looks like:

Host jumpbox
HostName (redacted)
User (redacted)
BatchMode yes
PasswordAuthentication no

Host *
ServerAliveInterval 60
TCPKeepAlive yes
#ProxyCommand ssh -q -A -t jumpbox nc %h %p # have tried
this one too
ProxyCommand ssh -q -t jumpbox ssh -p %p %h
ControlMaster auto
ControlPath ~/.ssh/ansible-%r@%h:%p
ControlPersist 8h
User (redacted)


Our boxes are running the standard Amazon AMI (which I am led to believe
are CentOS-based). We are open to rethinking the aforementioned strategy
re: the keys if that's the only way.

I'll be interested in anyone's thoughts on the matter. Thank you.

Best regards,

Andrew Merenbach
Senior DevOps Engineer at Edlio Inc

[a***@edlio.com]

Correction: the non-working command is "ssh -q -t jumpbox1 ssh -p %p %h"
(not awsjumpbox1)

Post by a***@edlio.com
Hi folks,
I'm encountering issues provisioning with Ansible via SSH from outside our
datacenter (well, AWS Virtual Private Cloud).
Our bastion hosts ("jump boxes") are configured to accept SSH keys from
staff members. Boxes behind the bastion hosts accept only the SSH keys
created for the jumpbox user on the bastion hosts. This makes our
administration easier in the event of key compromise staff turnover (only
need to rotate out on the jumpboxes).
The ProxyCommand examples I've seen for Ansible rely on agent forwarding
and netcat. I want Ansible to SSH in to the bastion host, then run a
second SSH command with the private key of the bastion host. Thus agent
forwarding is useless here. I've tried replacing the netcat ProxyCommand
with "ssh -q -t awsjumpbox1 ssh -p %p %h" and I get mysterious failures
SSH-2.0-OpenSSH_6.6.1_hpn13v11: command not found
It is sometimes useful to re-run the command using -vvvv, which prints SSH
debug output to help diagnose the issue.
[defaults]
host_key_checking = False
[ssh_connection]
ssh_args = -o ControlPersist=15m -F ssh.config -q
scp_if_ssh = True
Host jumpbox
HostName (redacted)
User (redacted)
BatchMode yes
PasswordAuthentication no
Host *
ServerAliveInterval 60
TCPKeepAlive yes
#ProxyCommand ssh -q -A -t jumpbox nc %h %p # have tried
this one too
ProxyCommand ssh -q -t jumpbox ssh -p %p %h
ControlMaster auto
ControlPersist 8h
User (redacted)
Our boxes are running the standard Amazon AMI (which I am led to believe
are CentOS-based). We are open to rethinking the aforementioned strategy
re: the keys if that's the only way.
I'll be interested in anyone's thoughts on the matter. Thank you.
Best regards,
Andrew Merenbach
Senior DevOps Engineer at Edlio Inc

[Paul Markham]

I had a play with this and I can't get it to work either.

Whilst running 'ssh -tt bastion01 ssh -tt server01' does works from the
command line, it doesn't work when used as a ProxyCommand in .ssh/config.

This makes sense though. When using ProxyCommand you're telling your ssh
client to connect to a remote server, the hops in between are transparent
to it. So, the private key needs to be on your local machine, not on some
intermediate machine.

If you're managing accounts and keys through Ansible, changing keys or
removing an account shouldn't be a big deal.