Commit Graph

102 Commits

Author SHA1 Message Date
Ole Markus With 466dcd001e Apply suggestions from code review
Co-authored-by: John Gardiner Myers <jgmyers@proofpoint.com>
2020-10-09 08:27:08 +02:00
Ole Markus With 809aa93634 Make use of kubelet service certificate 2020-10-09 08:27:08 +02:00
Ciprian Hacman 07ffd665a7 Allow container runtime to run before BootstrapKubeconfig 2020-09-12 08:13:40 +03:00
Ciprian Hacman 22ec1512dc Use numbers for distribution names 2020-08-17 07:25:43 +03:00
Ciprian Hacman e68ee80a93 Move and rename the "distros" package 2020-08-17 07:25:43 +03:00
Peter Rifel 4d9f0128a3
Upgrade to klog2
This splits up the kubernetes 1.19 PR to make it easier to keep up to date until we get it sorted out.
2020-08-16 20:56:48 -05:00
John Gardiner Myers fb381c4c8b Don't issue kubelet cert on masters before k8s 1.19 2020-08-15 10:30:21 -07:00
John Gardiner Myers c5871df319 Get kubelet certificate from kops-controller 2020-08-15 10:30:20 -07:00
John Gardiner Myers e405d24f8c Default kubelet authenticationTokenWebhook to true for k8s 1.19+ 2020-08-14 11:57:56 -07:00
John Gardiner Myers d2e7e2a41d Default kubelet authorization-mode to Webhook for k8s 1.19+ 2020-08-08 21:00:48 -07:00
Kubernetes Prow Robot 7a61e9f07a
Merge pull request #9403 from hakman/protokube-distroless
Use distroless image as base for Protokube
2020-07-12 20:32:34 -07:00
Justin SB 6cdf9d5001 Don't start kubelet in protokube
Previously as an optimization we would start the kubelet from
protokube, after we had mounted the disks.  This helped avoid e.g. the
apiserver going into backoff waiting for etcd.

However, this no longer achieves anything with etcd-manager - nothing
happens on this front until after we start the kubelet anyway.

Doing this both takes protokube out of the dependency sequence here
(slightly faster boot time), but also removes the systemd dependency
from the protokube image.  (So we can get a smaller image, perhaps
even distroless)
2020-07-05 14:41:29 +03:00
Ciprian Hacman 69511a998e Use kubelet docker-specific flags only for Docker 2020-07-05 07:57:10 +03:00
John Gardiner Myers 75ca231693 Move default machine type to NodeupConfig 2020-06-28 18:52:04 -07:00
John Gardiner Myers 44fb283e3f Move NodeLabels into the NodeupConfig 2020-06-28 18:52:03 -07:00
John Gardiner Myers 1ba0f0d463 Move Taints into the NodeupConfig 2020-06-28 18:51:42 -07:00
John Gardiner Myers 5e5f25703d Move KubeletConfig into the NodeupConfig 2020-06-28 18:51:16 -07:00
John Gardiner Myers 8a2dfeb377 Refactor buildMasterKubeletKubeconfig 2020-06-16 21:37:56 -07:00
Peter Rifel bc074e857c
Use ec2.DescribeInstanceTypes in awsup.GetMachineTypeInfo
This requires passing a cloud object in additional places throughout the validation package and originating mostly from cmd/kops

This means that some kops commands now require valid cloud provider credentials, but I don't think this is an issue because the vast majority of use-cases already require the same cloud provider credentials in order to interact with the state store.
2020-06-09 10:13:01 -05:00
ZouYu 2fc52ec6be fix some go-lint warning
Signed-off-by: ZouYu <zouy.fnst@cn.fujitsu.com>
2020-06-09 08:52:50 +08:00
Kubernetes Prow Robot d18e97140e
Merge pull request #9130 from johngmyers/pki-refactor
Refactor cert issuance code
2020-06-05 01:43:43 -07:00
John Gardiner Myers e88e0cf7ec Remove code supporting dropped k8s versions 2020-06-04 12:11:51 -07:00
John Gardiner Myers d64e760d5b Use pki.IssueCert() for nodeup client certs 2020-06-04 10:26:45 -07:00
John Gardiner Myers 2c02f0812c Upgrade docker/containerd/containeros hashes to SHA256 2020-05-29 21:12:36 -07:00
Kubernetes Prow Robot 6830cf6d44
Merge pull request #9065 from johngmyers/remove-distro
Remove support for CoreOS and Jessie
2020-05-27 23:22:01 -07:00
Ole Markus With d1ff25bb4e Remove some rather long networking nil checks 2020-05-22 08:08:58 +02:00
John Gardiner Myers 06c6ac1bee Remove support for CoreOS and Jessie 2020-05-04 23:09:52 -07:00
Ciprian Hacman 95ab317c8a Don't install static utils for Flatcar 2020-04-25 14:59:55 +03:00
Ciprian Hacman 17bb98b7d7 Add support for Ubuntu 20.04 (Focal) 2020-03-11 06:49:02 +02:00
Ciprian Hacman 162761cd64 Fix DNS loop on Ubuntu 18.04 (Bionic) 2020-01-16 18:25:19 +02:00
John Gardiner Myers 6e9dc8fc0f Remove code for unsupported k8s versions from nodeup 2020-01-12 19:30:34 -08:00
Ciprian Hacman d35977ac90 Update hardcoded dependencies on on docker.service 2019-12-29 04:43:40 +02:00
Kubernetes Prow Robot 4621df3c0a
Merge pull request #8155 from hakman/staticcheck-nodeup
nodeup/pkg/model - fix static check
2019-12-20 00:15:32 -08:00
Ciprian Hacman fa665943ac nodeup/pkg/model - fix static check 2019-12-18 22:07:31 +02:00
Ciprian Hacman ce9ed2e1fb Make sure Cluster.Spec.Containerd exists before checking the address 2019-12-16 15:26:52 +02:00
Ciprian Hacman 59ab0ca9d0 Pass containerd --address flag to kubelet if it exists 2019-12-16 00:59:26 +02:00
Ciprian Hacman 42b73b877d Add support for containerd container runtime 2019-12-15 21:37:57 +02:00
Salvatore Mazzarino f6908656bf Fix volume plugins dir
Signed-off-by: Salvatore Mazzarino <dev@mazzarino.cz>
2019-11-24 16:35:23 +01:00
Kubernetes Prow Robot 0c66a304c3
Merge pull request #7754 from tanjunchen/fix-up-gosimple-error
fix-up gosimple check error
2019-10-12 11:18:44 -07:00
liranp c6eb29a4b5
fix(nodeup): miscalculated `max-pods` when using amazon-vpc-cni 2019-10-10 12:02:12 +03:00
tanjunchen 43bf6bdab6 fix-up gosimple check error 2019-10-07 11:59:57 +08:00
Justin SB 728e582360
Fill out kops controller functionality
k8s 1.16 requires that we move label setting away from the kubelet, to
a central controller.  kops-controller is that controller.
2019-09-25 12:04:34 -04:00
mikesplain 9e55b8230a Update copyright notices
Also cleans some white spaces
2019-09-09 14:47:51 -04:00
Justin SB 8e2dbb6a99
Bootstrap: protokube labels its own node with node-role label
As of k8s 1.16, the node-role label is protected for security reasons.
We will introduce a controller to set those labels generically.
However, we need these labels to run the controller (only) on master
nodes.

To solve this bootstrapping problem, we use protokube to apply the
master role node labels to the master node only.  This isn't a
security problem because we assume that protokube on the master is
highly trusted - we are still administering labels centrally.

Then kops-controller can use this label to target the master nodes,
and run a central label controller.
2019-09-08 20:51:00 -04:00
Justin SB 810c65eb46
Copy well-known users from apiserver
This should eliminate the dependency on k8s.io/apiserver
2019-08-31 08:50:09 -07:00
Kashif Saadat 4514215656 Set and mount the correct volume plugin dir based on OS 2019-08-01 17:54:08 +01:00
Kubernetes Prow Robot a8e7109129
Merge pull request #7040 from spotinst/feature-spotinst-ocean
Spotinst: New instance group type: Ocean
2019-06-20 12:24:10 -07:00
liranp 80020052d3
feat(spotinst): new instance group type: ocean 2019-06-03 12:38:46 +03:00
Salvatore Mazzarino c7381f9a34 Flatcar support
Signed-off-by: Salvatore Mazzarino <dev@mazzarino.cz>
2019-05-31 12:20:27 +02:00
Justin SB 3e33ac7682
Change code from glog to klog
We don't call klog.InitFlags yet, because that will cause a flag
redefinition error until we get everyone to stop using glog.  That
will happen when we update to k8s 1.13.
2019-05-06 12:54:51 -04:00