Revert "Revert "Merge pull request #70 from riobard/refactor/vbx""

[riobard]

This reverts commit 0c4effd.

This seems to work now.

[SvenDowideit]

oh, and if no changes are needed (and that tests show it was just my build env was broken, then there are enough LGTM's already)

now for the testing - if @boot2docker/team-cli can build and test for each of the OS's and report, that would rock

[gmlewis]

OK, minor irritation on Mac OS X...

I ran:

• boot2docker-cli down
• boot2docker-cli delete (to kill my old one)
• boot2docker-cli init
• boot2docker-cli up

and during the up phase, I got this:

$ boot2docker-cli up
2014/03/24 20:45:40 Starting VM "boot2docker-vm"...
Waiting for VM "boot2docker-vm" to power on...
VM "boot2docker-vm" has been successfully started.
2014/03/24 20:45:40 Waiting for SSH server to start...
2014/03/24 20:45:59 Failed to connect to SSH port at localhost:2022: EOF

but right afterward, the ssh command succeeded. So I don't think we should have seen this error in such a short period (19 seconds), but maybe we can file an issue and investigate later.

[riobard]

@SvenDowideit I believe @schickling was trying to make the CI work, but was blocked by the TLS problem. See #13. I don't have a good solution right now, though :(

@gmlewis I encountered that problem too, which was why I added the retry mechanism here https://github.com/boot2docker/boot2docker-cli/pull/75/files#diff-488f601af322dcf4d74673602a7a67d8R154 Could you experiment with the value of n and the timeout delay on your machine? I set n = 10 and timeout to be 3 sec according to how it behaves on my machine, which is probably a bit too optimistic because my machine is pretty fast (it has a rather fast SSD), but I've always been wondering how it would work with slower machines.

[gmlewis]

@riobard On my MacBook Pro, it typically takes about 22 seconds for ssh to become available.

[SvenDowideit]

damn. I just got the EOF issue on master (ie not this change - we need to figure out something better.)

[riobard]

@gmlewis That's strange. the timeout and retry mechanism in this PR should tolerate up to 30 sec boot time in the VM. Did you experience the EOF error in the current master or in this PR? I can reliably reproduce the EOF error in the master, but not in this PR.

@SvenDowideit Can I just re-merge this one? I had 3 LGTMs before :)

[gmlewis]

Sorry, @riobard, it turns out that I was testing master. I will try again
to pull this PR and test it instead.