Mostly raising this issue to see if I can raise a clean PR against it.

for example
- ssh [email protected]

This problem was noticed when connecting to consumer ISP BT Broadband in the UK.

BT use a service backed by barefruit.co.uk which intercepts unresolvable DNS requests and returns a 'search' page with ads.

When connecting to this ISP and using a Macbook with VirtualBox, this appears to cause an issue with the S2I build (the nodejs-ex example fails during build). The S2I builder fails during git clone on git://openshift.example.com, unable to connect to the barefruit.co.uk IP returned by the BT router.

Is it possible to get the installation to default the domain on application creation to xxxx.cloudapps.example.com - currently it defaults to openshiftapps.com which if you forget to change it makes the service unreachable.

Is this a config option in OSE3 install?

Create an out of the box sample app (from a template) (UI and oc new-app)
Create an out of the box sample app (via a git repo) (UI and oc new-app)
Update app source -> push to PaaS
Connect Java debugger (optional)
oc rsh - remote shell to image (optional)
rollback (optional)
scalability (optional)

The openshift 3.0.1 offline image (and possibly others) have "crashkernel=auto" configured, which leads to RAM beeing reserved for kdump instead of beeing available for the kernel:

$ dmesg|grep crash
[ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-3.10.0-229.11.1.el7.x86_64 root=UUID=3e97515d-5a4a-418c-9754-252537d3ff45 ro crashkernel=auto LANG=en_US.UTF-8 net.ifnames=0
[ 0.000000] Reserving 161MB of memory at 704MB for crashkernel (System RAM: 4095MB)

This is probably not desired for this usage, the crashkernel option should be removed.

Phase 1 - get it working
Phase 2 - remove dependencies on VMDKstream.py and ovftool

@utherp0 FYI

The tomcat7/8 templates

- name: {APPLICATION_NAME}-tcp-8080

• name: tcp8080

(as system:admin)

oc edit templates/jws-tomcat7-basic-sti -n openshift

issue replicated in jws-tomcat8-basic-sti and possiblu others.

When a build fails it would be useful to pause to allow debugging in the current state.