Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Windows builds on VMWare Workstation failing on 1.1.2 #5622
Windows 2012 reboots are not detected properly on packer 1.1.2 with vmware-workstation (linux).
packer 1.0.4 and 1.1.1 work properly on both platforms.
I have reproduced this over 12 times with 100% consistency. I'm using the same packer definition that has been working for months with 1.0.4
Packer does not fail if using 1.0.4/1.1.1 OR if running 1.1.2 with OSX with vmware-fusion
This is easy, but very time consuming to reproduce, because it only happens after windows 2012 has completed installing windows updates, and then packer attempts to reboot the machine.
Watching the build process, windows always gets stuck shutting down after installing windows updates, and shows error. (Sorry for the picture, the ubuntu build machine doesn't have VNC).
After 5 to 30 minutes, the windows error always clears itself and windows reboots successfully. However packer 1.1.2 never detects that the vm has successfully rebooted, even with a 1 hour timeout.
Installed 1.15.7 of the vmware VIX api library
I've had to roll back to 1.0.4 or 1.1.1 to get windows 2012 images working again.
These are all the milestones from 1.1.2
What other information can I provide to help debug this? I want packer 1.1.2 to work with vmware workstation on linux.
I reproduced again and captured a more complete log
The first time packer tries to restart, it writes 0 bytes to stderr and stdout
Every time after that it writes
Also strange is that it only tried to restart the system 4 times. In many of the other times I've reproduced this it has tried to restart dozens of times.
Even more strange is that packer did not attempt to restart any more times from 9:31 until 10:30.
I also have this problem building Server 1709:
Apologies for the additional log prefixes... This is from a CI build under TeamCity. Same situation where the 1h timeout expires without any apparent attempt to retry the connection.
The build works just fine on my Linux desktop, which has packer 1.0.4.
referenced this issue
Dec 6, 2017
Edit: I'm talking rubbish, sorry..
a739623 is the commit that breaks things, but it's not fixed in Master as I suggested. I did something daft when trying to test that.
I managed to get it working in master by pulling
Thanks for going to all the work of that bisect. I made a PR reverting the change you think is the issue: #5672 While still keeping the code that was the main intent of the original PR associated with the bad commit. Can you try that patch out and let me know if it fixes things?