Commit Graph

326 Commits (582187b6fb2fab17565b50282350db672e94e250)

Author SHA1 Message Date
Kubernetes Submit Queue 582187b6fb Merge pull request #40758 from luxas/kubeadm_cleanup_rbac
Automatic merge from submit-queue

kubeadm: Move some code from apiclient.go to the dedicated apiconfig phase

**What this PR does / why we need it**:

Add constants and somewhat refactor the RBAC code as well + some cleanup.
I'm planning to rewrite the code in `setupmaster.go` later, but this PR has only the move of the code for easier reviewing.

**Special notes for your reviewer**:

This is broken out from: https://github.com/kubernetes/kubernetes/pull/40556

**Release note**:

```release-note
NONE
```

@mikedanese @pires @dmmcquay @dgoodwin
2017-02-01 13:20:37 -08:00
Lucas Käldström 79515ec880
Move some code from apiclient.go to the dedicated apiconfig phase package. Add constants and somewhat refactor the RBAC code as well 2017-02-01 19:06:51 +02:00
Paulo Pires ab1a5b53ab
kubeadm: set maxUnavailable and maxSurge for self-hosted component deployments.
Refs https://github.com/kubernetes/kubeadm/issues/127
2017-02-01 11:42:02 +00:00
Kubernetes Submit Queue 8ffada6699 Merge pull request #40698 from luxas/kubeadm_fix_authz_default
Automatic merge from submit-queue (batch tested with PRs 40707, 40698)

kubeadm: Default to control plane v1.6.0-alpha.1 and using RBAC

Also use constants for authz modes

**What this PR does / why we need it**:

Defaults to v1.6.0-alpha.1 (will be cut later today) because the certificates API group has been upgraded to beta, so `kubeadm join` at HEAD doesn't work on a `v1.5` cluster anyway.

By defaulting to v1.6.0-alpha.1, we can focus totally on v1.6 for kubeadm at HEAD, we don't support other versions in the upcoming v1.6 kubeadm release because of the alpha -> beta upgrades.

**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #

**Special notes for your reviewer**:

**Release note**:

```release-note
NONE
```

@mikedanese @pires
2017-01-31 03:29:44 -08:00
Kubernetes Submit Queue 41684eb22b Merge pull request #40628 from skriss/kubeadm_multi_endpoint_bugfix
Automatic merge from submit-queue (batch tested with PRs 40392, 39242, 40579, 40628, 40713)

fixed bug #36988 -- kubeadm join crashes when using multiple API endpoints

**What this PR does / why we need it**:
Simple bug fix for #36988 (kubeadm crashes when trying to join nodes to a master with multiple API endpoints) -- stopChan was getting closed multiple times, once per endpoint. Moved the close into the once.Do(...)

**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*:
fixes #36988 

**Special notes for your reviewer**:
this is my first kubernetes commit, just getting familiar with the process, so any/all feedback is appreciated

**Release note**:
2017-01-31 01:16:54 -08:00
Lucas Käldström fbc550c945
Default to control plane v1.6.0-alpha.1 for clusters deployed with kubeadm and using RBAC. Also use constants for authz modes 2017-01-31 09:24:15 +02:00
Kubernetes Submit Queue 1bc78add3e Merge pull request #40682 from dgoodwin/pod-security-context
Automatic merge from submit-queue (batch tested with PRs 38443, 40145, 40701, 40682)

Move kubeadm etcd SELinux options from container to pod.

**What this PR does / why we need it**:

Works around a bug that surfaces in Docker 1.12+ related to the pause
container's namespace and selinux labels being transferred to the etcd
container when it runs.

At present it appears that applying selinux options to a container may
be broken, or perhaps shouldn't be supported at all. Moving these to the
pod causes all containers (including pause) to run with the correct
labels.



**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #

**Special notes for your reviewer**:

Related to and partial fix for https://github.com/kubernetes/kubeadm/issues/107

This is one of several selinux related fixes in flight for upcoming releases, and newer versions of Docker. To successfully run kubeadm with selinux enforcing right now would like require a recent container-selinux build as uncovered in https://github.com/kubernetes/kubeadm/issues/107, a bugfix for the format labels in #40179, and finally this fix.

**Release note**:

```release-note
Fixed an SELinux issue in kubeadm on Docker 1.12+ by moving etcd SELinux options from container to pod.
```
2017-01-30 20:59:44 -08:00
deads2k c9a008dff3 move util/intstr to apimachinery 2017-01-30 12:46:59 -05:00
Devan Goodwin 65a48b57ff Move kubeadm etcd SELinux options from container to pod.
Works around a bug that surfaces in Docker 1.12+ related to the pause
container's namespace and selinux labels being transferred to the etcd
container when it runs.

At present it appears that applying selinux options to a container may
be broken, or perhaps shouldn't be supported at all. Moving these to the
pod causes all containers (including pause) to run with the correct
labels.
2017-01-30 10:16:22 -04:00
Dr. Stefan Schimanski 44ea6b3f30 Update generated files 2017-01-29 21:41:45 +01:00
Dr. Stefan Schimanski bc6fdd925d pkg/api/resource: move to apimachinery 2017-01-29 21:41:44 +01:00
Kubernetes Submit Queue babecf10dd Merge pull request #40194 from bruceauyeung/k8s-branch-fix-error-message-in-TestGetAddonImage
Automatic merge from submit-queue (batch tested with PRs 40132, 39302, 40194, 40619, 40601)

fix error message in TestGetAddonImage

Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>

**What this PR does / why we need it**:

1. fix incorrect error message in TestGetAddonImage
2. rename some variables to make them more reasonable
2017-01-27 18:32:24 -08:00
Kubernetes Submit Queue a812493aae Merge pull request #39302 from bruceauyeung/k8s-branch-fix-nil-check-and-typos
Automatic merge from submit-queue (batch tested with PRs 40132, 39302, 40194, 40619, 40601)

fix nil check and typos

**What this PR does / why we need it**:
1. nil error should be checked before defer statement.
2. fix some typos.

Signed-off-by: bruceauyeung ouyang.qinhua@zte.com.cn
2017-01-27 18:32:23 -08:00
Kubernetes Submit Queue 433e7a1fee Merge pull request #40132 from bruceauyeung/k8s-branch-add-invalid-and-valid-tokens-in-TestValidateToken-testcase
Automatic merge from submit-queue

add test tokens for TestValidateToken

Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>


**What this PR does / why we need it**:

1. add invalid tokens which token ID and token secret contain special characters such as `*`
2. add valid tokens which  token ID and token secret contain mixed lowercase characters and numeric numbers
2017-01-27 18:23:36 -08:00
Kubernetes Submit Queue 3ac0581a81 Merge pull request #39320 from bruceauyeung/k8s-branch-fix-incorrect-parameter-passing
Automatic merge from submit-queue

fix incorrect parameter passing

Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>


**What this PR does / why we need it**:
1. fix incorrect parameter passing when creating error
2. fix ineffectual assignment to err variable.
2017-01-27 15:50:54 -08:00
deads2k 2734f8f892 move dynamic and discovery clients 2017-01-26 08:37:06 -05:00
Kubernetes Submit Queue a931b7e238 Merge pull request #40357 from sttts/sttts-more-cutoffs-2
Automatic merge from submit-queue (batch tested with PRs 39538, 40188, 40357, 38214, 40195)

genericapiserver: cut off more dependencies – episode 2

Compare commit subjects.

approved based on #40363
2017-01-25 17:47:13 -08:00
Kubernetes Submit Queue 61d4a71853 Merge pull request #40429 from dgoodwin/self-hosted-wait
Automatic merge from submit-queue

Wait for the self-hosted control plane during kubeadm init.

**What this PR does / why we need it**:

Currently we are completing kubeadm init while the scheduler and
controller manager are likely still not yet up. In some cases if they
will fail, they won't ever come up.

Instead wait until each pod enters running state before exiting kubeadm.

**Release note**:

```release-note
NONE
```
2017-01-25 13:26:05 -08:00
Dr. Stefan Schimanski a0137e9b28 Update generated files 2017-01-25 19:49:45 +01:00
Dr. Stefan Schimanski d7eb3b6870 pkg/util: move uuid and strategicpatch into k8s.io/apimachinery 2017-01-25 19:45:09 +01:00
Devan Goodwin 1eb2e60dbf Wait for the self-hosted control plane during kubeadm init.
Currently we are completing kubeadm init while the scheduler and
controller manager are likely still not yet up. In some cases if they
will fail, they won't ever come up.

Instead wait until each pod enters running state before exiting kubeadm.
2017-01-25 12:16:03 -04:00
Timothy St. Clair bfe41927ba Cleaup Affinity post conversion from annotations to fields 2017-01-25 08:32:28 -06:00
Steve Kriss 29087066d0 fixed bug #36988 -- kubeadm crashes when using multiple API endpoints 2017-01-24 16:51:16 -08:00
Kubernetes Submit Queue df42444742 Merge pull request #40216 from sttts/sttts-more-cutoffs
Automatic merge from submit-queue (batch tested with PRs 39260, 40216, 40213, 40325, 40333)

genericapiserver: more dependency cutoffs

- cut-off pkg/api.Resource and friends - lgtm
- authn plugins -> k8s.io/apiserver - 
- webhook authz plugin -> k8s.io/apiserver - lgtm
- ~~pkg/cert -> k8s.io/apimachinery (will rebase on @deads2k's PR also moving it)~~
- split pkg/config into kubelet config merger and flags - lgtm
- split feature gate between generic apiserver and kube - lgtm
- move pkg/util/flag into k8s.io/apiserver - lgtm
2017-01-24 16:26:00 -08:00
Dr. Stefan Schimanski 2b8e938128 Update generated files 2017-01-24 20:56:03 +01:00
Dr. Stefan Schimanski 82826ec273 pkg/util/flag: move to k8s.io/apiserver 2017-01-24 20:56:03 +01:00
Clayton Coleman be6d2933df
refactor: Move *Options references to metav1 2017-01-24 13:41:51 -05:00
Clayton Coleman 7b1c715496
generated: Move references to metav1.*Options 2017-01-24 13:41:48 -05:00
Clayton Coleman fca7e00608
Move all *Options to metav1 2017-01-24 13:15:26 -05:00
deads2k 5a8f075197 move authoritative client-go utils out of pkg 2017-01-24 08:59:18 -05:00
Kubernetes Submit Queue 0fac54e65e Merge pull request #40316 from luxas/kubeadm_fix_apiserver_certs
Automatic merge from submit-queue (batch tested with PRs 40303, 40316)

kubeadm: Fix up the apiserver kubelet client cert

**What this PR does / why we need it**:

I didn't notice that the apiserver serving cert was _only_ a server cert, not both a server and client cert in https://github.com/kubernetes/kubernetes/pull/40292.

This PR fixes that misconception and generates a dedicated client cert for apiserver -> kubelet communication and harmonizes the `ca.crt` path to `/etc/kubernetes/pki/ca.crt` on both master and node.

**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*: fixes #

**Special notes for your reviewer**:

```console
$ systemctl cat kubelet
# /lib/systemd/system/kubelet.service
[Unit]
Description=kubelet: The Kubernetes Node Agent
Documentation=http://kubernetes.io/docs/

[Service]
ExecStart=/usr/bin/kubelet
Restart=always
StartLimitInterval=0
RestartSec=10

[Install]
WantedBy=multi-user.target

# /etc/systemd/system/kubelet.service.d/10-kubeadm.conf
[Service]
Environment="KUBELET_KUBECONFIG_ARGS=--kubeconfig=/etc/kubernetes/kubelet.conf **--authorization-mode=Webhook --client-ca-file=/etc/kubernetes/pki/ca.crt** --require-kubeconfig=true"
Environment="KUBELET_SYSTEM_PODS_ARGS=--pod-manifest-path=/etc/kubernetes/manifests --allow-privileged=true"
Environment="KUBELET_NETWORK_ARGS=--network-plugin=cni --cni-conf-dir=/etc/cni/net.d --cni-bin-dir=/opt/cni/bin"
Environment="KUBELET_DNS_ARGS=--cluster-dns=10.96.0.10 --cluster-domain=cluster.local"
ExecStart=
ExecStart=/usr/bin/kubelet $KUBELET_KUBECONFIG_ARGS $KUBELET_SYSTEM_PODS_ARGS $KUBELET_NETWORK_ARGS $KUBELET_DNS_ARGS $KUBELET_EXTRA_ARGS


$ curl -k https://localhost:10250/containerLogs/kube-system/weave-net-arm64-4t2xv/weave
Forbidden (user=system:anonymous, verb=get, resource=nodes, subresource=proxy)
$ curl --cert /etc/kubernetes/pki/apiserver-kubelet-client.crt --key /etc/kubernetes/pki/apiserver-kubelet-client.key -k https://localhost:10250/containerLogs/kube-system/weave-net-arm64-4t2xv/weave
[Works!!]
```

**Release note**:

```release-note
NONE
```

@liggitt @deads2k
2017-01-24 01:57:11 -08:00
bruceauyeung 21e49646b8 fix incorrect parameter passing
Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>
2017-01-24 14:37:57 +08:00
Clayton Coleman 9009c1ac14
generated: informer,client 2017-01-23 17:52:47 -05:00
Clayton Coleman 469df12038
refactor: move ListOptions references to metav1 2017-01-23 17:52:46 -05:00
Clayton Coleman f623a8c908
Make all groups use metav1.AddToGroupVersion for ListOptions 2017-01-23 16:53:18 -05:00
Lucas Käldström 695fb7440e
Fix up the misunderstanding about the apiserver kubelet client cert 2017-01-23 21:45:48 +02:00
Kubernetes Submit Queue 5ea80569ae Merge pull request #40227 from deads2k/client-11-easy
Automatic merge from submit-queue

make client-go tools/auth and tools/clientcmd authoritative

These moved easily.  Pure mechanical.  I'll have a couple snip pulls for some of the more complicated packages.  `pkg/util/homedir` was a transitive that moved with. It was only used in `kubectl` and you'd need it to be consistent with `clientcmd` loading order.

@sttts
2017-01-23 09:07:17 -08:00
Kubernetes Submit Queue 071844e35f Merge pull request #40292 from luxas/kubeadm_node_ca
Automatic merge from submit-queue (batch tested with PRs 38445, 40292)

kubeadm: Secure apiserver -> kubelet communication and set storage backend to etcd3

**What this PR does / why we need it**:

Switch storage backend to etcd3
Writes ca.crt down to disk so we can set `--client-ca-file=/etc/kubernetes/ca.crt` for the kubelet.
Adds --kubelet-client-{certificate,key} to the apiserver args and make it allowed to access the kubelets
In some cases the `path` dependency is switched to `filepath`

**Which issue this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close that issue when PR gets merged)*:

fixes https://github.com/kubernetes/kubeadm/issues/118
fixes https://github.com/kubernetes/kubeadm/issues/129

**Special notes for your reviewer**:

This PR is should make it possible to secure the apiserver -> kubelet communication.

**Release note**:

```release-note
NONE
```

@pires @mikedanese @andrewrynhard @liggitt @deads2k
2017-01-23 08:11:08 -08:00
deads2k 335ef747d6 mechanical 2017-01-23 10:03:55 -05:00
deads2k 7cc9702402 separate announce factories 2017-01-23 08:02:14 -05:00
Lucas Käldström 6a37f450ae
Add --kubelet-client-{certificate,key} to the apiserver args and make it allowed to access the kubelets 2017-01-23 13:16:04 +02:00
Lucas Käldström 6579c94559
Write ca.crt down to disk in kubeadm join 2017-01-23 13:15:19 +02:00
Lucas Käldström 22ce0fdf2d
Set apiserver's storage backend to etcd3 2017-01-23 12:08:01 +02:00
Lucas Käldström 9e7ef10f95
Don't require the kubeconfigs and certs to not exist; that's now handled in later in the individual phase code. Also fix a small test 2017-01-21 01:00:17 +02:00
Lucas Käldström 13499f443a
Generate two certs and two private keys; only the necessary ones; make the certs and kubeconfig phases work with valid files already on-disk and some cleanup 2017-01-21 00:33:06 +02:00
Lucas Käldström 741b0b8c9f
Make the discovery deployment load the CA from a file 2017-01-21 00:30:35 +02:00
Kubernetes Submit Queue 0efee9a67e Merge pull request #40152 from liggitt/sign-cert-org
Automatic merge from submit-queue (batch tested with PRs 40187, 40231, 40152)

Update client/server cert generation utilities

Limit generated cert usage to client or server use.
2017-01-20 13:29:48 -08:00
Kubernetes Submit Queue b5929bfb2b Merge pull request #38789 from jessfraz/cleanup-temp-dirs
Automatic merge from submit-queue (batch tested with PRs 37228, 40146, 40075, 38789, 40189)

Cleanup temp dirs

So funny story my /tmp ran out of space running the unit tests so I am cleaning up all the temp dirs we create.
2017-01-20 12:34:58 -08:00
Kubernetes Submit Queue 7f9997a456 Merge pull request #40075 from apprenda/kubeadm_self_hosted
Automatic merge from submit-queue (batch tested with PRs 37228, 40146, 40075, 38789, 40189)

kubeadm: add optional self-hosted deployment

**What this PR does / why we need it**: add an optional self-hosted deployment type to `kubeadm`, for master components only, namely `apiserver`, `controller-manager` and `scheduler`.

**Which issue this PR fixes**: closes #38407

**Special notes for your reviewer**: /cc @aaronlevy @luxas @dgoodwin

**Release note**:
```release-note
kubeadm: add optional self-hosted deployment for apiserver, controller-manager and scheduler.
```
2017-01-20 12:34:56 -08:00
Paulo Pires 724ce6a8a5
kubeadm: add temporary --self-hosted flag. 2017-01-20 17:41:47 +00:00