July 19, 2017

Experiments with Vagrant and Packer On Fedora

  —A look at Vagrant and Packer (ancient Fedora).

I've had a strange experience with Packer on Fedora 13 (yes, 13). This may prove interesting to anyone encountered the following error from Packer:
==> virtualbox-iso: Error waiting for SSH: ssh: handshake failed: ssh: unable to authenticate, attempted methods [none password keyboard-interactive], no supported methods remain
I'm using Packer v0.10.1.

I found that adjusting ssh_timeout did not seem to have any effect. I ended up setting ssh_handshake_attempts equal to 100.

What I learned in investigating this problem is this is the error message you get when your handshake attempts threshold exceeded. This message implies an authentication problem, but in my case it occurred because SSH wasn't running yet.

To remedy this, confirm whether it's an SSH problem or not by executing:
PACKER_LOG=1 PACKER_LOG_PATH=out.log packer build --debug TEMPLATE
Stop at the step:
==> virtualbox-iso: Pausing after run of step 'StepTypeBootCommand'. Press enter to continue.
Run Packer on a template with headless set to false (the default) so you can watch the virtual machine boot. Login once things are up and running. If the install completes you can also "Press enter to continue". If Packer connects over SSH you can rule out an SSH problem.

To confirm whether it's problem with your Packer template check ssh_wait_timeout and ssh_timeoutssh_wait_timeout is deprecated in favour of ssh_timeout.  Curiously, neither seemed to have any effect on Fedora 13. At one point, I set my timeout to hours and watched Packer shutdown the virtual machine because it couldn't connect is the space of a few minutes.

I was successful using both parameters on CentOS 7.0 and 7.2 as well as Debian 7.11.0 and 8.5.0. 

The problem doesn't appear to be my Packer template (Fedora uses the same template as CentOS and Debian). 

It isn't Packer (it works properly on CentOS and Debian).

Weird.
comments powered by Disqus