Tarmac is timing out when test runner integration test( test_run_tests) is included

[Para Siva]

With tarmac command timeout being a hard coded 900 seconds, including test_test_runner in the engine makes tarmac runs timeout out with
'Command appears to be hung. There has been no output for 900 seconds. Sending SIGTERM'

Running test_test_runner test in the local deployment takes roughly 525 secs. This itself is significant enough to cause the tarmac to timeout with 900s.

Also what happens in the test runner instance inside the tarmac environment (which is essentially an lxc container in an lcy02 instance) is that running tests with even a small package waits (kind of forever) with this message:

 'Not enough random bytes available. Please do some other work to give the OS a chance to collect more entropy! (Need 286 more bytes)'

This entropy issue is not seen in runs on lxc deployments on actual hardware, i.e. locally on our machines. Not seen on hpcloud deployments either.

It appears that haveged (Installed: 1.1-2 in tr instance) should take care of this entropy issue and it appears, it does, in the local deployments on actual hw and on a non lxc deployment of hpcloud.

[Para Siva (psivaa)]

Just to be clear, I think if we are to include this test, we need to
1. Fix this entropy issue in env similar to tarmac
2. Increase the tarmac timeout

[Vincent Ladeuil (vila)]

Better tracked on https://app.asana.com/0/8312550429076/14926029353148