no error code for puppet execution

[Alexander Tivelkov]

Deploying two-node non-HA environment on virtual machines (Host - MacOS X Mavericks, virtualization in VirtualBox 4.3.2, using virtual box scripts to prepare the environment). OS - cents, networking - neutron on VLANs.

Controller deployment fails with errors, which indicate that puppet didn't return any return code at all. For example
(/Stage[main]/Horizon/Exec[horizon_compress_styles]) Failed to call refresh: python manage.py compress returned instead of one of [0] at /etc/puppet/modules/horizon/manifests/init.pp:212

See attached puppet logs for details.

The same deployment succeeds if run in different virtualisation system (VMWare fusion instead of VirtualBox)

[Sergey Vasilenko (xenolog)]

It's a strange Puppet heisenbug.
Puppet lost return code of script start.

[Nastya Urlapova (aurlapova)]

Alexander, please, specify iso version.

[Alexander Tivelkov (ativelkov)]

I observed this problem on both Fuel-4.0-67-2013-11-22_12-58-41.iso and Fuel-4.0-74-2013-11-25_14_36_04.iso

[Dmitry Ilyin (idv1985)]

Please try this patch if you can reproduce the bug https://github.com/Mirantis/fuel/pull/887/files

[Mike Scherbakov (mihgen)]

Is this applicable for both 3.2-fixes & master branches?

[Dmitry Ilyin (idv1985)]

Yes, for both

[Vladimir Kuklin (vkuklin)]

Spent several deployments in order to reproduce it - still no result