Error installing coreos image: [ERROR] openpgp: invalid data: user ID packet not followed by self-signature

[rkuzsma]

I freshly installed the latest corectl and kube-cluster-osx today on OS X El Capitan 10.11.2 (15C50). When I click Setup->Initial setup of Kube-Cluster VMs, I get this error:

[ERROR] openpgp: invalid data: user ID packet not followed by self-signature

Full console log:

22:17:53-myusername~$ /Applications/Kube-Cluster.app/Contents/Resources/first-init.command; exit;

Setting up Kubernetes Cluster for macOS

Reading ssh key from /Users/myusername/.ssh/id_rsa.pub

/Users/myusername/.ssh/id_rsa.pub found, updating configuration files ...

Set CoreOS Release Channel:
 1)  Alpha (may not always function properly)
 2)  Beta
 3)  Stable (recommended)

Select an option: 3


Please type Nodes RAM size in GBs followed by [ENTER]:
[default is 2]:
Changing Nodes RAM to 2GB...


Creating 5GB sparse disk (QCow2) for Master ...
-
Created 5GB Data disk for Master

Please type Nodes Data disk size in GBs followed by [ENTER]:
[default is 15]:

Creating 15GB sparse disk (QCow2) for Node1...
-
Created 15GB Data disk for Node1

Creating 15GB sparse disk (QCow2) for Node2...
-
Created 15GB Data disk for Node2

Starting k8smaster-01 VM ...
> booting k8smaster-01 (1/1)
[WARN] /Users/myusername/.coreos/images/stable/1122.2.0/coreos_production_pxe.vmlinuz missing - stable/1122.2.0 ignored
---> downloading and verifying stable/1122.2.0
[ERROR] openpgp: invalid data: user ID packet not followed by self-signature

Master VM has not booted, please check '~/kube-cluster/logs/master_vm_up.log' and report the problem !!!

Press [Enter] key to continue...

[rkuzsma]

Found a workaround:

If I open a new iterm window and run this line of code to install the VMs, I don't get the openpgp error:

$ ~/bin/corectl load settings/k8smaster-01.toml
> booting k8smaster-01 (1/1)
[WARN] /Users/myusername/.coreos/images/stable/1122.2.0/coreos_production_pxe.vmlinuz missing - stable/1122.2.0 ignored
---> downloading and verifying stable/1122.2.0
32.63 MB / 32.63 MB [==================================================================================================================================] 100.00 %
---> SHA512 hash for coreos_production_pxe.vmlinuz OK
220.78 MB / 220.78 MB [================================================================================================================================] 100.00 %
---> SHA512 hash for coreos_production_pxe_image.cpio.gz OK
---> stable/1122.2.0 ready
---> 'k8smaster-01' started successfuly with address 192.168.64.2 and PID 87297
---> 'k8smaster-01' boot logs can be found at '/Users/myusername/.coreos/running/7E79BC00-7E3C-4B19-9CDF-83F99FD73A23/log'
---> 'k8smaster-01' console can be found at '/Users/myusername/.coreos/running/7E79BC00-7E3C-4B19-9CDF-83F99FD73A23/tty'

I can then manually execute all the code inside the first-init.command shell script successfully to get everything up and running.

[rimusz]

that's is a very weird issue, it could be that first time the download got interrupted.
@AntonioMeireles would be possible to make corectl tool to try to re-download the missing fail automatically?

[rkuzsma]

It might have been an interrupted download, but I am not so sure... After the issue occurred, I removed my ~/kube-cluster directory, re-ran the installation, and got the same error. The error happened immediately, before the download progress meter appears. If other folks arent seeing this, though, then maybe consider it a connectivity problem on my end.