Rubygem-Staypuft: While deploying, during the OS installation phase - there's no timeout on failure. Environment: ISO: rhelosp-installer-live-20140617.1.poodle.iso ruby193-rubygem-foreman_openstack_simplify-0.0.6-7.el6ost.noarch openstack-foreman-installer-2.0.6-1.el6ost.noarch openstack-puppet-modules-2014.1-14.5.el6ost.noarch foreman-installer-staypuft-0.0.19-1.el6ost.noarch ruby193-rubygem-staypuft-0.1.3-1.el6ost.noarch Steps to reproduce: 1. Don't configure (or misconfigure) the installation media for RHEL. 2. Start the deployment. 3. The console of the node shows "Couldn't find kernel image..." and the boot process has stopped. Results: The deployment progress shows "State: running", and the host progress is 15% forever. Expected: The user should be notified, that something is wrong - requiring intrevention.
This goes along with the other usability bugs around deployment status. Petr, anything we can do here? I'm nervous about a timeout since a slow network or complex kickstart could influence the time dramatically as could various other thins like CDN speed.
We can add a timeout and make it configurable in Settings UI. What time would you put there as a default?
1 hr? I really don't know. I could argue that the provision shouldn't take longer than 15 min, but I think that might be too aggressive for some people.
We could also use configurable time out for other deployment tasks. There are times for example when a task fails pretty early but is waiting for a time out. This can result in the task looping continuously until the timeout value is reached. If we could configured the task time to the environment we can get feedback to the end user more quickly.
Don't believe this is a blocker, setting exception.
This is resolved in the latest releases