Commit Graph

61 Commits

Author SHA1 Message Date
Alexander Block
fd8e6d03e6 Fix test failures in manifests_test.go 2016-12-05 10:28:23 +01:00
Alexander Block
8373ab5890 kubeadm: Let apiserver and controller-manager host-mount /etc/pki when required
fixes: #36150
2016-12-05 10:11:15 +01:00
Clayton Coleman
3454a8d52c refactor: update bazel, codec, and gofmt 2016-12-03 19:10:53 -05:00
Clayton Coleman
5df8cc39c9 refactor: generated 2016-12-03 19:10:46 -05:00
Kubernetes Submit Queue
00e369b14f Merge pull request #37813 from kubernetes/revert-36625-branch-eliminate-recursive-call-attemptToUpdateMasterRoleLabelsAndTaints
Automatic merge from submit-queue (batch tested with PRs 35300, 36709, 37643, 37813, 37697)

Revert "[kubeadm] use iteration instead of recursion in function"

Reverts kubernetes/kubernetes#36625

Removing the recursive call means that `n` is never updated, so you never succeed in the update, and you've creating an infinite loop.

Also, this entire bit of functionality should be a patch and you won't have to worry about conflicts.  

@luxas
2016-12-03 08:55:59 -08:00
Derek McQuay
b3a56abed0 kubeadm: updated pkg node tests to correct type
updated ""k8s.io/kubernetes/pkg/api" to apiv1 "k8s.io/kubernetes/pkg/api/v1" to fix broken types in tests.
2016-12-02 08:42:46 -08:00
Derek McQuay
644a0ceec9 kubeadm: adding test owner and bazel update 2016-12-02 08:42:46 -08:00
Derek McQuay
49b53b8644 kubeadm: unit tests for app/master/apiclient.go 2016-12-02 08:42:46 -08:00
Derek McQuay
ada63282ed kubeadm: unit tests for app/master/addons.go 2016-12-02 08:42:45 -08:00
Derek McQuay
a54515d281 kubeadm: unit tests for app/master/discovery.go 2016-12-02 08:42:45 -08:00
Derek McQuay
fe73a8a8ff kubeadm: unit tests for app/master/pki.go 2016-12-02 08:42:45 -08:00
Derek McQuay
bd4d98fabe kubeadm: unit tests for app/master/manifests.go 2016-12-02 08:42:45 -08:00
Derek McQuay
124dab6c5a kubeadm: unit tests for app/master/kubeconfig.go 2016-12-02 08:42:45 -08:00
Derek McQuay
cb3b08a8d4 kubeadm: unit tests for app/master/tokens.go 2016-12-02 08:42:45 -08:00
Kubernetes Submit Queue
9a67c20b3d Merge pull request #37327 from jasonbrooks/pr-kubeadm-selinux
Automatic merge from submit-queue

change unconfined_t to spc_t

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

When installing kube via kubeadm on a system w/ selinux enabled, it's necessary to disable selinux in order for the etcd and kube-discovery containers to run. 

The kube etcd and discovery pods are currently set to unconfined_t in order to avoid disabling selinux, but the correct type for an unconfined container is spc_t. For more information, see http://danwalsh.livejournal.com/2016/10/03/.
2016-12-02 01:00:42 -08:00
David Eads
ce7b08af92 Revert "[kubeadm] use iteration instead of recursion in function" 2016-12-01 08:48:45 -05:00
Kubernetes Submit Queue
aef15aa875 Merge pull request #36625 from bruceauyeung/branch-eliminate-recursive-call-attemptToUpdateMasterRoleLabelsAndTaints
Automatic merge from submit-queue

[kubeadm] use iteration instead of recursion in function

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

before this PR, function `attemptToUpdateMasterRoleLabelsAndTaints` recursively call itself. there are some defeats in it:
1. potential stack overflow.
2. unnecessary extra  `json.Marshal` calls.
3. unnecessary extra `client.Nodes().List` calls.

this PR rewrite `attemptToUpdateMasterRoleLabelsAndTaints` function, use iterate instead of recursion.
so these 3 defeats the metioned above are gone.
Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>
2016-12-01 00:39:14 -08:00
Kubernetes Submit Queue
c74fae7b14 Merge pull request #36172 from apprenda/kubeadm-tests-table-driven
Automatic merge from submit-queue

kubeadm: changed tests to be table driven

Small change to migrate tests in kubeadm/app/util/tokens_test.go to be table driven which should make adding more tests in the future easier and also have them match how other tests are being written. 

This PR is part of the ongoing effort to add tests (#35025)
2016-12-01 00:39:07 -08:00
Chao Xu
bcc783c594 run hack/update-all.sh 2016-11-23 15:53:09 -08:00
Chao Xu
e3cf383181 cmd/kubeadm 2016-11-23 15:53:09 -08:00
Jason Brooks
bf153fc1d3 change unconfined_t to spc_t
The kube etcd and discovery pods are set to unconfined_t in
order to avoid disabling selinux, but the correct type for
an unconfined container is spc_t. For more information, see
http://danwalsh.livejournal.com/2016/10/03/.
2016-11-22 16:25:31 -08:00
Derek McQuay
bff1af69e9 kubeadm: fixed typo in kubeadm/app/master/pki.go 2016-11-15 09:55:27 -08:00
bruceauyeung
f3c5a6ef8f eliminate recursive call attemptToUpdateMasterRoleLabelsAndTaints
Signed-off-by: bruceauyeung <ouyang.qinhua@zte.com.cn>
2016-11-11 12:25:36 +08:00
Kubernetes Submit Queue
15fa0df93e Merge pull request #35975 from justinsb/role_labels_2
Automatic merge from submit-queue

Add constants for node role labels
2016-11-07 15:02:58 -08:00
Justin Santa Barbara
cef8315ae8 Add constants for node role labels
Rather than sharing well-known strings, we should be declaring these in
the API.
2016-11-04 18:00:35 -04:00
Mike Danese
a00db0a32f kubeadm: move kubeadm over to apigroup defaulting 2016-11-01 16:48:39 -07:00
Paulo Pires
eb6eeb704a kubeadm: gofmt'ed cmd/kubeadm. 2016-10-29 11:38:53 -04:00
Derek McQuay
b7c685d421 kubeadm: EnvParam struct and GlobalEnvParam
Previously, GetEnvParams (now called SetEnvParams) had no way of being altered unless
it was through enviroment variables. These changes allow for a global
EnvParam to be set and also altered while still initally getting their value from
set enviroment variables. This change is especially helpful for testing
(see kubeadm/app/util/kubeconfig_test.go).
2016-10-29 09:41:51 -04:00
Kubernetes Submit Queue
ffdfe9fa9b Merge pull request #35119 from errordeveloper/tidy-up-kubeadm
Automatic merge from submit-queue

Start tidying kubeadm up

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

This PR addresses https://github.com/kubernetes/kubernetes/pull/33262#discussion_r80337511.

**Release note**:
```release-note
NONE
```
2016-10-24 02:26:36 -07:00
Mike Danese
3b6a067afc autogenerated 2016-10-21 17:32:32 -07:00
Ilya Dmitrichenko
9703df391f Eliminate half-baked multi-architecture support, but keep node affinity architecture-agnostic (fix #33916) 2016-10-21 14:56:02 +01:00
Ilya Dmitrichenko
b7b0822654 Refactor getComponentCommand() for readability 2016-10-19 14:56:55 +01:00
Ilya Dmitrichenko
6f57775669 Ensure MasterConfiguration is refered to as cfg throughout 2016-10-17 12:08:11 +01:00
Ilya Dmitrichenko
327dec43fb Add flags for alternative API and discovery ports (close #34311 #34307 #33638) 2016-10-17 12:08:11 +01:00
Lucas Käldström
4ac49a74aa Remove duplicate --etcd-servers arguments in kubeadm 2016-10-16 21:48:16 +03:00
Kubernetes Submit Queue
4a223efd27 Merge pull request #34573 from errordeveloper/fix-early-deployment-issue
Automatic merge from submit-queue

Test API more extensivelly before declaring readiness 

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

It's possible that first deployment kubeadm creates will hit `deployments.extensions "kube-discovery" is forbidden: not yet ready to handle request`, which comes from NamespaceLifecycle admission controller and has something to do with cache. According to @derekwaynecarr, we need to create a namespace-scoped resource to really check for this. I didn't want to make a check with deployment of whatever comes first right now, and decided to have explicit step for this in `apiclient.go`.

**Which issue this PR fixes**: fixes #34411

**Special notes for your reviewer**: @kubernetes/sig-cluster-lifecycle 

**Release note**:
```release-note
NONE
```
2016-10-13 01:44:24 -07:00
Ilya Dmitrichenko
0cb54e7eb2 Append first address from --api-advertise-addresses to kube-apiserver flags 2016-10-12 12:59:24 +01:00
Ilya Dmitrichenko
53e393ad42 Test API more extensivelly before declaring readiness (close #34411) 2016-10-11 22:33:39 +01:00
Kubernetes Submit Queue
851705ea59 Merge pull request #34147 from mikedanese/kubeadm-api
Automatic merge from submit-queue

kubeadm: turn api into a real apigroup

@kubernetes/sig-cluster-lifecycle
2016-10-10 09:59:54 -07:00
Mike Danese
45b92085c4 kubeadm: still run cm if not pod cidr is specified 2016-10-07 12:31:10 -07:00
Mike Danese
b3dae78c62 kubeadm: turn api into a real apigroup 2016-10-07 11:14:53 -07:00
Mike Danese
56ea178e7c kubeadm: refactor config
1) break object into substructures
2) seperate a config object for master and node
2016-10-03 14:44:18 -07:00
Kubernetes Submit Queue
64d2b12d21 Merge pull request #33859 from luxas/not_so_aggressve_probes
Automatic merge from submit-queue

Do not probe so aggressively which may lead to unnecessary restarts

@errordeveloper @mikedanese PTAL

I came across a case where etcd restarted about 5-10 times because the load was very high on the machine. 
The load seems to have lead to that the `etcd` container occasionally didn't respond to the probe, which caused many restart and made the whole thing even worse.

Maybe we should remove the etcd probe totally? I don't know, what do you think?
This is at least a try to loosen the limits here...
2016-10-03 05:55:42 -07:00
Kubernetes Submit Queue
347d448180 Merge pull request #33831 from rustyrobot/print-cert-info
Automatic merge from submit-queue

Kubeadm: print information about certificates

Prints basic information about certificates to the user.

Example of `kubeadm init` output:
```
<master/pki> generated Certificate Authority key and certificate:
Issuer: CN=kubernetes | Subject: CN=kubernetes | CA: true
Not before: 2016-09-30 11:19:19 +0000 UTC Not After: 2026-09-28 11:19:19 +0000 UTC
Public: /etc/kubernetes/pki/ca-pub.pem
Private: /etc/kubernetes/pki/ca-key.pem
Cert: /etc/kubernetes/pki/ca.pem
<master/pki> generated API Server key and certificate:
Issuer: CN=kubernetes | Subject: CN=kube-apiserver | CA: false
Not before: 2016-09-30 11:19:19 +0000 UTC Not After: 2017-09-30 11:19:19 +0000 UTC
Alternate Names: [172.18.76.239 10.0.0.1 kubernetes kubernetes.default kubernetes.default.svc kubernetes.default.svc.cluster.local]
Public: /etc/kubernetes/pki/apiserver-pub.pem
Private: /etc/kubernetes/pki/apiserver-key.pem
Cert: /etc/kubernetes/pki/apiserver.pem
<master/pki> generated Service Account Signing keys:
Public: /etc/kubernetes/pki/sa-pub.pem
Private: /etc/kubernetes/pki/sa-key.pem
```

Example of `kubeadm join` command:
```
<node/csr> received signed certificate from the API server:
Issuer: CN=kubernetes | Subject: CN=system:node:minion | CA: false
Not before: 2016-09-30 11:28:00 +0000 UTC Not After: 2017-09-30 11:28:00 +0000 UTC
```

Fixes #33642
cc @kubernetes/sig-cluster-lifecycle
2016-10-01 11:31:25 -07:00
Lucas Käldström
60274778bc Set FailureTreshold to 8 for kubeadm components in order to not restart unless really necessary 2016-10-01 16:25:27 +03:00
Evgeny L
12f8c979ba kubeadm: user-friendly certificates formatting 2016-10-01 04:57:30 +00:00
Zach Aller
bd1d93e2e0 fix kubeadm on AWS so that kube-controller has access to certs for amazon api 2016-09-28 18:47:23 +00:00
Evgeny L
8f586d916e Kubeadm: fix SELinux rules for kubernetes discovery service 2016-09-27 13:24:29 +00:00
Lucas Käldström
b17e107def Various improvements plus added a version command 2016-09-25 22:38:39 +03:00
Ilya Dmitrichenko
a023085a5f Address comments in review 2016-09-24 17:28:34 +01:00