Automatic merge from submit-queue
make: allow overriding quick-release opts
Prior to this change, `make quick-release KUBE_RELEASE_RUN_TESTS=y`
would still not run the tests. This change makes it so such overriding
works.
It also specifies the options as variables, which helps some shells with
tab-completion.
Finally, I plan to add a new "quick-release" option that not everyone will want on in #35249 (`RELEASE_HYPERKUBE_ONLY`), and this sets the needed groundwork to make such an "opt-in for quickness" flag possible.
**Release note**:
```release-note
NONE
```
Automatic merge from submit-queue (batch tested with PRs 41649, 41658, 41266, 41371, 41626)
Allow `build/run.sh make help` to run
**What this PR does / why we need it**: typo fix; `build/run.sh make help` doesn't work for me on OS X w/ docker 1.13.x, this PR fixes that
**Release note**:
```release-note
NONE
```
/cc @spxtr
This commit wires-in the pre-existing `--container-runtime` flag for
local node_e2e testing.
This is needed in order to further skip docker specific testing
and validation.
Local CRI node_e2e can now be performed via
`make test-e2e-node RUNTIME=remote REMOTE=false`
which will also take care of passing the appropriate arguments to
the kubelet.
This makes "make bazel-build" no longer build docker images and
release tars, which are the slowest and least incremental parts. They
will still be built in CI since we run //:ci-artifacts.
Prior to this change, `make quick-release KUBE_RELEASE_RUN_TESTS=y`
would still not run the tests. This change makes it so such overriding
works.
It also specifies the options as variables, which helps some shells with
tab-completion.
Automatic merge from submit-queue
New command: "kubeadm token generate"
As part of #33930, this PR adds a new top-level command to kubeadm to just generate a token for use with the init/join commands. Otherwise, users are left to either figure out how to generate a token on their own, or let `kubeadm init` generate a token, capture and parse the output, and then use that token for `kubeadm join`.
At this point, I was hoping for feedback on the CLI experience, and then I can add tests. I spoke with @mikedanese and he didn't like the original propose of `kubeadm util generate-token`, so here are the runners up:
```
$ kubeadm generate-token # <--- current implementation
$ kubeadm generate token # in case kubeadm might generate other things in the future?
$ kubeadm init --generate-token # possibly as a subcommand of an existing one
```
Currently, the output is simply the token on one line without any padding/formatting:
```
$ kubeadm generate-token
1087fd.722b60cdd39b1a5f
```
CC: @kubernetes/sig-cluster-lifecycle
**Release note**:
<!-- Steps to write your release note:
1. Use the release-note-* labels to set the release note state (if you have access)
2. Enter your extended release note in the below block; leaving it blank means using the PR title as the release note. If no release note is required, just write `NONE`.
-->
``` release-note
New kubeadm command: generate-token
```
The KUBE_VERBOSE environment variable sets the verbosity level to
use. Log messages can specify a verbosity by setting the V
variable. e.g.
V=2 kube::log::info foo bar
Would only print "foo bar" if $KUBE_VERBOSE >= 2.
Automatic merge from submit-queue
Fix make verify
@thockin this fixes `make verify` which was previously failing due to undefined `BRANCH` and `VERBOSE` variables, ptal
Automatic merge from submit-queue
Makefile fixes and speed
A few fixes that, together, make the 'make' experience smoother and more seamless.
* `make clean all` works (@soltysh)
* `make clean` does no spend 5 seconds loading deps just to remove them
* deps building is faster
Fixes#28890
Ref #8830
Previously even 'make clean' would first produce dependency information (slow)
and sinclude all the metadata, possibly restarting the make, before removing
the things it just did.
Now dependencies are only processed when something explicitly depends on them,
by calling 'make' on the rules around generated files, rather than sincluding
them.
Net result: make clean is faster and safer. Additionally, bash tab-completion
for 'make' does not run the slow 'find' any more.
Mike Danese