Premature timeout when waiting for WinRM for be ready

[zl4bv]

When waiting for the WinRM transport to be ready, a timeout occurs which causes test kitchen to crash.

$ kitchen test -l DEBUG
-----> Starting Kitchen (v1.4.0)
D      Berksfile found at /home/ben/tktest/Berksfile, loading Berkshelf
D      Berkshelf 3.2.4 library loaded
D      Winrm Transport requested, loading WinRM::Transport gem (~> 1.0)
D      WinRM::Transport 1.0.1 library loaded
-----> Cleaning up any prior instances of <default-windows2012r2>
-----> Destroying <default-windows2012r2>...
       Finished destroying <default-windows2012r2> (0m0.00s).
-----> Testing <default-windows2012r2>
-----> Creating <default-windows2012r2>...
       Creating <>...
If you are not using an account that qualifies under the AWS
free-tier, you may be charged to run these suites. The charge
should be minimal, but neither Test Kitchen nor its maintainers
are responsible for your incurred costs.

D      Creating EC2 Instance..
       Instance <i-3a78e3e6> requested.
       EC2 instance <i-3a78e3e6> created.
       Waited 0/300s for instance <i-3a78e3e6> to become ready.
       Waited 5/300s for instance <i-3a78e3e6> to become ready.
       Waited 10/300s for instance <i-3a78e3e6> to become ready.
       Waited 15/300s for instance <i-3a78e3e6> to become ready.
       Waited 20/300s for instance <i-3a78e3e6> to become ready.
       Waited 25/300s for instance <i-3a78e3e6> to become ready.
       EC2 instance <i-3a78e3e6> ready.
D      [WinRM] opening remote shell on plaintext::http://ec2-52-64-46-171.ap-southeast-2.compute.amazonaws.com:5985/wsman<{:disable_sspi=>true, :basic_auth_only=>true, :user=>"kitchen", :pass=>"K1tch3nR0ck5"}>
>>>>>> ------Exception-------
>>>>>> Class: Kitchen::ActionFailed
>>>>>> Message: Failed to complete #create action: [execution expired]
>>>>>> ----------------------
>>>>>> Please see .kitchen/logs/kitchen.log for more details
>>>>>> Also try running `kitchen diagnose --all` for configuration

After the above error has occurred I can manually wait for WinRM to be ready, type kitchen converge and it will happily continue.

I would expect the WinRM transport to respect the transport.connection_timeout and transport.max_wait_until_ready settings in the first instance.

Also it looks like receive timeout has recently been made configurable in the winrm library - I'm wondering if it would be helpful to make connect timeouts configurable there too.

More details on the issue here: https://gist.github.com/zl4bv/ed5a1b11ecdc610518d2

[tyler-ball]

Hey @zl4bv, thanks for the bug report. Kitchen-ec2 currently doesn't support Windows machines - #96 tracks our last attempt to add this support in.

The next release of kitchen-ec2 (0.10.0) will add back in this Windows support. I'm currently working on it and will address all your bugs before I release 0.10.0

[zl4bv]

Hi @tyler-ball, Kitchen-ec2 support for Windows as of v0.9.1 isn't as broken/far away as you might think. Last night (NZ time) I found that the only issues that have prevented me from using Kitchen-ec2 with Windows are as follows:

• WinRM not enabled by default on EC2 instances so have to manually enable it with user-data
• This issue - have to wait for WinRM to be enabled; WinRM transport currently too impatient
• Default username for WinRM transport doesn't work (issue Transport defaults windows username to ./administrator test-kitchen#688)
• Command to create EC2 JSON for Ohai doesn't work because the command is targeted at *NIX-based machines - which breaks the end of the kitchen 'create' phase

With these issues resolved (custom user-data, PR test-kitchen/test-kitchen#728, and PR #134) I was able to start a Windows Server 2012 R2 Base instance in EC2 and converge a basic Chocolatey install recipe using Chef Solo.

The would-be-nice features for Windows support that I can for the immediate future are:

• WinRM enabled automatically by Kitchen-ec2 (could be salvaged from Creating windows-guest-support branch for kitchen-ec2 #96)
• EC2-generated Administrator password (Add feature to get Windows password generated by AWS #133)
• Windows AMIs in amis.json
• Store generated admin password and pass it to RDP command when running kitchen login

I can see a lot of good stuff in #96 but it seems that it will become increasingly harder to merge the changes as master has evolved a lot since that PR was originally opened - and I feel this will delay any development of Windows support. I'm quite keen to see Windows support soon, hence the PRs I opened to fix issues with Windows support with kitchen-ec2 as it stands in the master branch today.

[tyler-ball]

@zl4bv thanks for the awesome breakdown - I'm working on pulling together you're multiple PRs into one and testing it this week. I'll certainly appreciate your review when I have it together!

[tyler-ball]

Closing this as test-kitchen/test-kitchen#728 was merged and released as Test Kitchen 1.4.1

[ohdihe]

Still having this issue ec2 windows. It freezes at ec2 ready but the instance is launched I can see it via the terminal. When I press control c to get my prompt back and enter kitchen list, it say last action not created.
Changed the ami-id on the .kitchen.yml file to a linux instance and everything seems okay. Don't know why it only happens with window ec2.

[dragon788]

@ohdihe We had an issue where somebody had specified a Linux custom user_data.sh script at the top level and it was attempting to launch this script on Windows nodes as well and hanging them. Can you open a new issue and paste your minimal/anonymized .kitchen.yml ?