NFD

Ashlesh, how do you managing images? Just in plain virtualbox?

I have suggestion to invest time into building vagrant images and use puppet as a configuration management. It would still be virtualbox, but this would allow simpler instantiation and management of vms. Moreover, others would be able to easily instantiate VM that is used for testing.

Yes I am using VirtualBox command line to make VMs. Right now I just made three VMs. How do I build vagrant images?

I have limited knowledge about vagrant, as I basically just started learning. Some tutorial is available at https://docs.vagrantup.com/v2/boxes.html.

I can only say that creating "box" in vagrant is straightforward and simple. Most likely, you will be able to reuse the virtualbox image you already created.

Another piece that I briefly mentioned is Puppet. I have not used it at all, but a lot of projects use it for management purposes. So we need to investigate how it can help us.

Alright I have installed all dependencies on the three machines. Now should I install a puppet agent on them and make a Vagrant base box?

go ahead if you know how to do those.

Hi Alex

I am confused about installing puppet server. Should I just install puppet agent on each machine and you will install and configure a server or should I install and configure a puppet server?

I'm not sure. I don't really know what puppet is and how to use it. You can install agents on VMs and then try to figure out how to use them (and then tell us).

And, just in case. It is optional. If it is too complicated to use / there are no clear benefits, then don't spend too much time on it.

You can try to upload them to vagrant cloud (https://vagrantcloud.com). I haven't done it before, but they have some documentation about it.

CentOS and Fedora required additional tweaking of /etc/sudoers: default config requires tty to run sudo, which prevents jenkins script to do any meaningful task.

See http://unix.stackexchange.com/questions/122616/why-do-i-need-a-tty-to-run-sudo-if-i-can-sudo-without-a-password

Also failed and also most likely because of limited memory:

• http://jenkins.named-data.net/job/ndn-cxx/OS=CentOS7/1689/console

I disabled requiretty for jenkins user.

Still failing at random places. http://jenkins.named-data.net/job/ndn-cxx/OS=CentOS7/1696/consoleFull

I have no idea what is going on and how to fix that... There is almost 2Gb free space available plus 2Gb of swap (which is not used).

I notice that both CentOS 7 slaves are deployed on the same site. One needs to be moved to fulfill resiliency requirements and prevent downtime due to single site failure.