-
Notifications
You must be signed in to change notification settings - Fork 40k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
REPLACED BY #6983 - Fix #2741. Add support for alternate Vagrant providers: ... #6879
Conversation
…, VMWare Workstation, and Parallels.
Thanks for your pull request. It looks like this may be your first contribution to a Google open source project, in which case you'll need to sign a Contributor License Agreement (CLA). 📝 Please visit https://cla.developers.google.com/ to sign. Once you've signed, please reply here (e.g.
|
I signed it! |
CLAs look good, thanks! |
@derekwaynecarr you touched Vagrantfile once. Are you a good reviewer, or can you suggest one? |
I can review tomorrow. Thanks for the enhancement. Sent from my iPhone
|
… specifying a raw URL.
Ignore this comment. Original and updated content preserved below. For some reason, I don't have any visibility into the Shippable failure from my latest commits. I haven't been able to reproduce any failures on my end with 9fca1cf or 50d80de (but I can back them out if necessary). The idea was to be able to support specific versions for boxes hosted (e.g.) via vagrantcloud.com. UPDATE # 1: Scratch that, I was able to re-run the failing Shippable tests directly from my fork/branch (copy of logs). Forgive me (I'm not that familiar with Go, or Shippable for that matter), but I'm not sure I understand the error. It looks like there are bunch of missing package-related failures during integration testing that have nothing to do with my latest check-in? UPDATE # 2: I reverted back to 468fb89 in a separate branch (posita/kubernetes@vmware-parallels-support-2741-shippable-test), and that failed (even though the exact same commit passed less than 6 hours prior). UPDATE # 3: I merged posita/kubernetes@vmware-parallels-support-2741-shippable-test with GoogleCloudPlatform/kubernetes@master and it still failed. Then I ran posita/kubernetes@master, and even that failed, so I'm pretty sure it's not me? Either that, or it's something with my own Shippable account configuration and I'm not seeing the same error? 😕 |
…KUBERNETES_{MASTER,MINION}_MEMORY environment variables (with reversion to KUBERNETES_MEMORY).
…rantfile` so that `vagrant ssh minion-N` actually works (possibly related to #4633?). Collapse memory override fallback to more easily allow different default values for master/minion (if that ever becomes desirable).
Shippable is flaky, your PR is safe from Shippable. I am trying to resolve a few other flaky issues with the Vagrant setup right now based on other changes that happened on the system to run more resources in pods, but once that is stabilized I will look to merge this. |
…provisioning (see Parallels/vagrant-parallels#171, including my comment).
…T_VAGRANT_PROVIDER` ➡️ `VAGRANT_DEFAULT_PROVIDER`. (Oops! 😧).
UPDATE: This PR is replaced by #6983 and should not be merged.
...VMWare Fusion, VMWare Workstation, and Parallels. (Tagging #2741.)
I tried to mirror the Vagrant philosophy and do the right thing in most cases without tweaking configuration (unless you really wanted to override something). The fall-back order is:
If any of these are installed properly (including any required Vagrant plugins),
KUBERNETES_PROVIDER=vagrant .../cluster/kube-up.sh
should just work. 😁 _CAVEAT__: I do not have a license to VMWare Fusion. I'm pretty sure it will function, but I have only tested Parallels._To override, you can set
DEFAULT_VAGRANT_PROVIDER
(e.g., if you have both VMWare and Parallels installed, but want to use Parallels):You can override the box (by name):
And even specify a version:
Or specify a URL for the box itself, in which case you now must provide your own name:
WARNING: This breaks the existing behavior of
KUBERNETES_BOX_URL
, which is now ignored unlessKUBERNETES_BOX_NAME
is also set. Previously that name was fixed for everything (default or specified by URL) asfedora20
, irrespective of underlying OS/version. Now if a default box is used, the name is set internally (currently onlykube-fedora20
), but users who specify their own box URL are now forced to name it. To avoid clashes or confusion, it is recommended (but not enforced) that the name distinguish the box from the defaults (e.g.,posita-fedora21
,kickass-custom-centos
, etc.).This PR also introduces the ability to set the master memory size independently of the minions (it is included with this PR instead of a separate one because memory configuration is Vagrant provider-specific; see, e.g., this vs. this):
KUBERNETES_MEMORY
is maintained for backward compatibility. The following are equivalent to the previous example: