Compare commits

..

58 Commits

Author SHA1 Message Date
Ganesh Maharaj Mahalingam 0e280f40fa cri-o update fixes
ClearLinux now has cri-o version 1.14.1 which allows multiple plugin
locations and also creates /opt/cni/bin by default. We no longer need
the hacks for them

Fixes: #82
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-06-04 11:53:36 -07:00
Ganesh Maharaj Mahalingam d31d78c193 calico-node fails to come up cause of all.rp_filter
Currently net.ipv4.conf.all.rp_filter is set to 2 in Clear and
calico-node fails to come up unless that value is either 0 or 1.

Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-05-30 14:17:44 -07:00
Jose Carlos Venegas Munoz 04487a2cfe reset_stack: stop containerd and crio
- If stop CRI service if is running
- Restart the CRI service only if enabled
- Do not enable crio on reset_stack.sh this should be only part of the
setup.

Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
2019-05-29 18:20:33 -07:00
Jose Carlos Venegas Munoz e67630b0bd create_stack: quote variables to avoid code expands
General fixes make my vim mark less warnings from shellcheck

Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
2019-05-24 11:40:49 -07:00
Jose Carlos Venegas Munoz e08c92f6d5 stack: do not provide CRI socket
kubeadm autodetects the socket path based on defaults from well known
CRI servers.

Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
2019-05-24 11:37:08 -07:00
Jose Carlos Venegas Munoz d00a5ea9d8 README: update flavor setup information.
Add docuementation to use ./clr-k8s-examples/reset_stack.sh help

Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
2019-05-23 20:38:45 -07:00
Jose Carlos Venegas Munoz a9b3b5c506 create_stack: Add init and cni subcommands
Add more subcommands to to increase granularity

init: start cluster
cni: setup network

This functionality already existed, we only handle
subcommands in a more dynamic way.

Fixes: #92

Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
2019-05-23 18:56:59 -07:00
Ganesh Maharaj Mahalingam 24f248b02f Set reverse path forwarding to strict.
Calico requires the default reverse path forwarding to be either 0 (no
validation) or 1 (strict validation). The default value of 2 (loose
validation) prevents calico from completing the setup and the pod is
always stuck getting ready.

Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-05-18 14:13:13 -07:00
Ganesh Maharaj Mahalingam d688cbb693 Download OVMF if it doesn't exist
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-05-01 09:57:04 -07:00
Ganesh Maharaj Mahalingam 4e82cf411d Fix readme to add k8s version dependency
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-04-04 10:56:23 -07:00
Saikrishna Edupuganti af56d51499 Remove applying runtimeclass CRD
CRD was removed as part of 1.14 update. Missed update to create_stack

Fixes: #84

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-04-04 10:01:36 -07:00
Eric Ernst f863bb7f1f runtimeClass updates for kubernetes 1.14
RuntimeClass moved from alpha to beta. Remove the feature gates, remove
the runtimeClass CRD since it is a built in type, and take the new data
structure into account.

Fixes: #77

Signed-off-by: Eric Ernst <eric.ernst@intel.com>
2019-04-03 11:23:40 -07:00
Eric Ernst d0c7bb3513 kata-deploy: update to match what's on Kata master
We updated to use containerd v2 shim - pull in latest from master

Signed-off-by: Eric Ernst <eric.ernst@intel.com>
2019-03-28 13:09:39 -07:00
Saikrishna Edupuganti 4f38e14245 Turn on static policy for CPU manager
This is finally supported in Kata 1.6
Update memory limit for kata to run in guaranteed test case

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-03-27 10:27:12 -07:00
Saikrishna Edupuganti 3b7e7f75d1 Updated top-level README to point to admit-kata
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-03-27 10:26:54 -07:00
Saikrishna Edupuganti 39b454045d Copied the README too
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-03-27 10:26:54 -07:00
Saikrishna Edupuganti fe57aa93ff Upstream artifact for kata admission controller
Get upstream artifacts of kata admission controller from
https://github.com/kata-containers/tests/tree/master/kata-webhook
Commit: 5ad2cec

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-03-27 10:26:54 -07:00
Saikrishna Edupuganti 0417be1cdd Update README.md
Add note about environment variables that can be used to customize the setup
2019-03-26 16:35:25 -07:00
Ganesh Maharaj Mahalingam 02750e991c Make firecracker setup configurable
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-03-22 17:11:57 -07:00
Saikrishna Edupuganti dc345d4bdf Do not restart containerd, remove stop kubelet
In case of containerd there is no need to restart. Restart for crio is needed
when it manages the networks ns lifecycle. Removed the kubelet stop step which
was a leftover from earlier device plugin install.

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-03-20 13:23:16 -07:00
Manohar Castelino 935bb69cfd Eliminate legacy annotation
We only support kubernetes versions with support for runtimeClass.
Remove the legacy annotations to reduce confusion.

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-03-14 10:03:37 -07:00
Manohar Castelino e5c55c3c0d Update webhook setup and documentation
The admission controller is now able to access full context.
Hence explicit tagging of pod manifests to exclude them from
kata is no longer required.

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-03-14 10:03:37 -07:00
Manohar Castelino a6d2f487d8 Disable proxy
Disable proxy setup if the host does not have proxy setup in
the enviornment variables

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-03-12 09:57:44 -07:00
Saikrishna Edupuganti ed9f71ee92 Update README.md 2019-03-10 09:23:50 -07:00
Saikrishna Edupuganti c2e932ac7b Hotfix for rsync update in new clear (box?)
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-26 17:31:55 -08:00
Saikrishna Edupuganti 13226c0c72 Update README.md 2019-02-22 19:35:41 -08:00
Saikrishna Edupuganti be17bc3ab7 Use kata-deploy to install and configure kata
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-22 19:34:49 -08:00
Saikrishna Edupuganti b76ff0a4d3 Update multi-network readme
Add steps to show use of helper scripts
2019-02-20 17:05:24 -08:00
Ganesh Maharaj Mahalingam d57b6b187f kata_firecracker.sh script cleanup.
* Remove duplicate lifecycle entry in crio.conf
* Remove fc config.toml as that is provided by the distro.

Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-02-20 14:46:00 -08:00
Ganesh Maharaj Mahalingam b92dd0e5ed Check crio.conf before copying
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-02-19 13:02:40 -08:00
Manohar Castelino 5cd217a510 Disable devicemapper udev sync
Disable devicemapper udev sync. Without this the storage device
will not be discovered post system reboot

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-02-12 19:11:41 -08:00
Manohar Castelino ac79ef38c5 Downloading OVMF causes issues with libvirt
The parent box is already downloading the OVMF firmware.
Just use the firmware that is already downloaded.

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-02-12 16:17:18 -08:00
Saikrishna Edupuganti a43f9bd3de Update to latest metrics-server 0.3.x works
With kubelet service file fixed and correct args set in
metrics-server-deployment.yaml, autoscale seems to be working reliably.

Fixes: #5

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-08 14:06:18 -08:00
Saikrishna Edupuganti b925d25730 Re-organize tests,Add resource/limits/runtimeclass
Currently cpumanagerpolicy:static is partially broken.
Hope these files catch future regressions

Added a script to generate yamls for different runtimeclass from
a template with different resource constraints on the pod

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-08 09:43:53 -08:00
Saikrishna Edupuganti f238ed6db1 Update README.md 2019-02-05 18:22:04 -08:00
Saikrishna 88b6d23aa9 Simplify sriov.sh, do not generate random MAC
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-05 18:22:04 -08:00
Saikrishna 7805f2c942 Show both netdevice and vfio examples
Addressing comments -
1a. added flag to `sriov.sh` to indicate bind to vfio-pci
1b. systemd unit sets up two pools 1 per mode
2a. updated `sriov-conf.yaml` to use the two pools
3a. included network examples for both resource pools
3b. included pod examples for both resource pools

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-05 18:22:04 -08:00
Saikrishna fc99db500b Enable sriov-network-device-plugin vfio mode
Adding a cni to mirror MAC address of VF on a veth pair with results
from ipam applied to enable DPDK apps to configure themselves when
operating sriov-network-device-plugin in vfio mode

Updated helper sriov.sh to allow for binding VFs to vfio-pci on boot

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-05 18:22:04 -08:00
Saikrishna Edupuganti 72659b3ba3 Make script executable and qualify run as pod
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-02-05 17:58:30 -08:00
Manohar Castelino c234cca3c3 kata-webhook: Change location to katadocker
The kata-webhook is now hosted within the Kata repository.
Switch to using the kata generated images

https://cloud.docker.com/u/katadocker/repository/docker/katadocker/kata-webhook-example
https://github.com/kata-containers/tests/tree/master/kata-webhook

Signed-off-by: Manohar Castelino <manohar.r.castelino@intel.com>
2019-02-01 15:43:34 -08:00
Saikrishna Edupuganti a979f0056b Revert "kata 1.5 supports cpumanagerpolicy static"
This reverts commit c8006ca944.

Got too excited. Lets try this in next release
2019-01-31 21:24:20 +00:00
Ganesh Maharaj Mahalingam f8dc5cd815 Change disk image names
Avoid creating unique names for disks as vagrant-libvirt today does not
delete those images when VMs are deleted. Once that is fixed we can move
back to using unique names. Until then this should suffice.

Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-01-31 13:34:35 -08:00
Ganesh Maharaj Mahalingam f33ecc1091 Switch vagrant boxes
Switching boxes to the one maintained by Antonio at
https://github.com/AntonioMeireles/ClearLinux-packer. They now support
libvirt and will eventually be part of the clearlinux tree.

Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-01-30 10:18:22 -08:00
Saikrishna Edupuganti 97f8ea2506 Moving around to allow early exit
In case we need to exit early this would have copied the credentials
into the home dir

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-30 09:10:52 -08:00
Saikrishna Edupuganti c8006ca944 kata 1.5 supports cpumanagerpolicy static
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-30 09:10:52 -08:00
Saikrishna Edupuganti 0d01668513 Update readme with location of files and cleanup
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-17 14:14:33 -08:00
Saikrishna Edupuganti f5b1aa7ee5 Mount the working dir at vagrant user home
Instead of mounting at /vagrant it will now be at /home/vagrant/clr-k8s-examples

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-17 14:14:33 -08:00
Ganesh Maharaj Mahalingam 5beadf6a23 Enable kubectl autocompletion by default in the VMs
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
2019-01-14 19:17:55 -08:00
Saikrishna 42ffa8df71 Auto-generate multus conf. Do not exit sriov DS
Instead of relying on configmap to be created/customized by the user to
generate multus conf, we autogenerate it using the current default CNI
conf file on the host. This way CNI's with host specific conf details
will continue to work.

```
{
  "type": "calico",
  "nodename": "clr-01",
   ...
}
```

Since SR-IOV device may not be present on all nodes, the daemonset
should not exit as this will put the pod in restart loop. This would
cause the kubelet and runtimes to repeatedly restart, as per
initContainer logic.

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-08 15:34:45 -08:00
Saikrishna 39f3e063d3 Example systemd scripts to setup SR-IOV on nodes
Signed-off-by Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-07 11:41:40 -08:00
Saikrishna aecda4bf65 Add README for multus and sriov
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-07 11:41:40 -08:00
Saikrishna e974d2fc51 Add Dockerfile to generate the image with multus, sriov binaries
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-07 11:41:40 -08:00
Saikrishna a0261d6507 Adding multus and sriov artifacts
This set of yamls installs multus-cni, sriov-cni and sriov-device-plugin

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-07 11:41:40 -08:00
Saikrishna e4ea313ff2 Adding gitignore
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-04 22:04:15 -08:00
Saikrishna Edupuganti 3291e1582d Rename test-deploy-kata-fire.yaml to test-deploy-kata-fc.yaml
Typo
2019-01-04 14:44:25 -08:00
Saikrishna bfa29b5543 [README] Update to reflect filename changes
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-04 14:32:00 -08:00
Saikrishna f96b48beca Make naming consistent, kata points to kata-qemu
Runtime class kata-qemu created and older kata point to kata-qemu
handler. Now all the handlers match the crio.conf settings.

Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
2019-01-04 14:32:00 -08:00
Eric Ernst eff1fc9021 update runtimeClass, binary naming
Signed-off-by: Eric Ernst <eric.ernst@intel.com>
2019-01-04 14:32:00 -08:00
60 changed files with 1252 additions and 357 deletions
+3
View File
@@ -0,0 +1,3 @@
.vagrant
OVMF.fd
@@ -0,0 +1,12 @@
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: system:aggregated-metrics-reader
labels:
rbac.authorization.k8s.io/aggregate-to-view: "true"
rbac.authorization.k8s.io/aggregate-to-edit: "true"
rbac.authorization.k8s.io/aggregate-to-admin: "true"
rules:
- apiGroups: ["metrics.k8s.io"]
resources: ["pods"]
verbs: ["get", "list", "watch"]
@@ -1,3 +1,4 @@
---
apiVersion: rbac.authorization.k8s.io/v1beta1
kind: ClusterRoleBinding
metadata:
@@ -1,3 +1,4 @@
---
apiVersion: rbac.authorization.k8s.io/v1beta1
kind: RoleBinding
metadata:
@@ -1,3 +1,4 @@
---
apiVersion: apiregistration.k8s.io/v1beta1
kind: APIService
metadata:
@@ -1,3 +1,4 @@
---
apiVersion: v1
kind: ServiceAccount
metadata:
@@ -22,10 +23,19 @@ spec:
k8s-app: metrics-server
spec:
serviceAccountName: metrics-server
volumes:
# mount in tmp so we can safely use from-scratch images and/or read-only containers
- name: tmp-dir
emptyDir: {}
containers:
- name: metrics-server
image: gcr.io/google_containers/metrics-server-amd64:v0.2.0
image: k8s.gcr.io/metrics-server-amd64:v0.3.1
imagePullPolicy: Always
command:
- /metrics-server
- --source=kubernetes.summary_api:https://kubernetes.default.svc?kubeletHttps=true&kubeletPort=10250&useServiceAccount=true&insecure=true
args:
- --logtostderr
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP,Hostname,ExternalIP
volumeMounts:
- name: tmp-dir
mountPath: /tmp
@@ -1,3 +1,4 @@
---
apiVersion: v1
kind: Service
metadata:
@@ -1,3 +1,4 @@
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
@@ -9,15 +10,6 @@ rules:
- pods
- nodes
- nodes/stats
- namespaces
verbs:
- get
- list
- watch
- apiGroups:
- "extensions"
resources:
- deployments
verbs:
- get
- list
+1 -1
View File
@@ -1 +1 @@
f90c6705d2381ea2db1a6343da6c400bd2ef4cb2
92d8412788e27ee669d38f21f20bad5342211884
@@ -2,8 +2,6 @@ apiVersion: v1
kind: Namespace
metadata:
name: rook-ceph-system
labels:
kata: "false"
---
apiVersion: apiextensions.k8s.io/v1beta1
kind: CustomResourceDefinition
-2
View File
@@ -2,8 +2,6 @@ apiVersion: v1
kind: Namespace
metadata:
name: rook-ceph
labels:
kata: "false"
---
apiVersion: v1
kind: ServiceAccount
@@ -0,0 +1,46 @@
---
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: kubelet-kata-cleanup
namespace: kube-system
spec:
selector:
matchLabels:
name: kubelet-kata-cleanup
template:
metadata:
labels:
name: kubelet-kata-cleanup
spec:
serviceAccountName: kata-label-node
nodeSelector:
katacontainers.io/kata-runtime: cleanup
containers:
- name: kube-kata-cleanup
image: katadocker/kata-deploy
imagePullPolicy: Always
command: [ "bash", "-c", "/opt/kata-artifacts/scripts/kata-deploy.sh reset" ]
env:
- name: NODE_NAME
valueFrom:
fieldRef:
fieldPath: spec.nodeName
securityContext:
privileged: false
volumeMounts:
- name: dbus
mountPath: /var/run/dbus
- name: systemd
mountPath: /run/systemd
volumes:
- name: dbus
hostPath:
path: /var/run/dbus
- name: systemd
hostPath:
path: /run/systemd
updateStrategy:
rollingUpdate:
maxUnavailable: 1
type: RollingUpdate
@@ -0,0 +1,69 @@
---
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: kata-deploy
namespace: kube-system
spec:
selector:
matchLabels:
name: kata-deploy
template:
metadata:
labels:
name: kata-deploy
spec:
serviceAccountName: kata-label-node
containers:
- name: kube-kata
image: katadocker/kata-deploy
imagePullPolicy: Always
lifecycle:
preStop:
exec:
command: ["bash", "-c", "/opt/kata-artifacts/scripts/kata-deploy.sh cleanup"]
command: [ "bash", "-c", "/opt/kata-artifacts/scripts/kata-deploy.sh install" ]
env:
- name: NODE_NAME
valueFrom:
fieldRef:
fieldPath: spec.nodeName
securityContext:
privileged: false
volumeMounts:
- name: crio-conf
mountPath: /etc/crio/
- name: containerd-conf
mountPath: /etc/containerd/
- name: kata-artifacts
mountPath: /opt/kata/
- name: dbus
mountPath: /var/run/dbus
- name: systemd
mountPath: /run/systemd
- name: local-bin
mountPath: /usr/local/bin/
volumes:
- name: crio-conf
hostPath:
path: /etc/crio/
- name: containerd-conf
hostPath:
path: /etc/containerd/
- name: kata-artifacts
hostPath:
path: /opt/kata/
type: DirectoryOrCreate
- name: dbus
hostPath:
path: /var/run/dbus
- name: systemd
hostPath:
path: /run/systemd
- name: local-bin
hostPath:
path: /usr/local/bin/
updateStrategy:
rollingUpdate:
maxUnavailable: 1
type: RollingUpdate
@@ -0,0 +1,29 @@
---
apiVersion: v1
kind: ServiceAccount
metadata:
name: kata-label-node
namespace: kube-system
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: node-labeler
rules:
- apiGroups: [""]
resources: ["nodes"]
verbs: ["get", "patch"]
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: kata-label-node-rb
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: node-labeler
subjects:
- kind: ServiceAccount
name: kata-label-node
namespace: kube-system
@@ -1,6 +0,0 @@
kind: RuntimeClass
apiVersion: node.k8s.io/v1alpha1
metadata:
name: fire
spec:
runtimeHandler: fire
@@ -0,0 +1,5 @@
kind: RuntimeClass
apiVersion: node.k8s.io/v1beta1
metadata:
name: kata-fc
handler: kata-fc
@@ -0,0 +1,12 @@
---
kind: RuntimeClass
apiVersion: node.k8s.io/v1beta1
metadata:
name: kata
handler: kata
---
kind: RuntimeClass
apiVersion: node.k8s.io/v1beta1
metadata:
name: kata-qemu
handler: kata-qemu
@@ -1,6 +0,0 @@
kind: RuntimeClass
apiVersion: node.k8s.io/v1alpha1
metadata:
name: kata
spec:
runtimeHandler: kata
@@ -1,26 +0,0 @@
kind: CustomResourceDefinition
apiVersion: apiextensions.k8s.io/v1beta1
metadata:
name: runtimeclasses.node.k8s.io
labels:
addonmanager.kubernetes.io/mode: Reconcile
spec:
group: node.k8s.io
version: v1alpha1
versions:
- name: v1alpha1
served: true
storage: true
names:
plural: runtimeclasses
singular: runtimeclass
kind: RuntimeClass
scope: Cluster
validation:
openAPIV3Schema:
properties:
spec:
properties:
runtimeHandler:
type: string
pattern: '^([a-z0-9]([-a-z0-9]*[a-z0-9])?(\.[a-z0-9]([-a-z0-9]*[a-z0-9])?)*)?$'
@@ -0,0 +1,33 @@
# Build multus plugin
FROM golang:1.10 AS multus
RUN git clone -q --depth 1 https://github.com/intel/multus-cni.git /go/src/github.com/intel/multus-cni
WORKDIR /go/src/github.com/intel/multus-cni
RUN ./build
# Build sriov plugin
FROM golang:1.10 AS sriov-cni
RUN git clone -q -b dev/k8s-deviceid-model https://github.com/Intel-Corp/sriov-cni.git /go/src/github.com/intel-corp/sriov-cni
WORKDIR /go/src/github.com/intel-corp/sriov-cni
RUN ./build
# Build sriov device plugin
FROM golang:1.10 AS sriov-dp
RUN git clone -q https://github.com/intel/sriov-network-device-plugin.git /go/src/github.com/intel/sriov-network-device-plugin
WORKDIR /go/src/github.com/intel/sriov-network-device-plugin
RUN make
# Build vfioveth plugin
FROM busybox as vfioveth
RUN wget -O /bin/jq https://github.com/stedolan/jq/releases/download/jq-1.6/jq-linux64
COPY cni/vfioveth /bin/vfioveth
RUN chmod +x /bin/vfioveth /bin/jq
# Final image
FROM centos/systemd
WORKDIR /tmp/cni/bin
COPY --from=multus /go/src/github.com/intel/multus-cni/bin/multus .
COPY --from=sriov-cni /go/src/github.com/intel-corp/sriov-cni/bin/sriov .
COPY --from=vfioveth /bin/vfioveth .
COPY --from=vfioveth /bin/jq .
WORKDIR /usr/bin
COPY --from=sriov-dp /go/src/github.com/intel/sriov-network-device-plugin/build/sriovdp .
@@ -0,0 +1,76 @@
# Multi-Network
## Daemonset
We launch a `Daemonset` with an `initContainer` which sets up the CNI
directories on the host with the necessary binaries and configuration files.
> NOTE: SR-IOV devices are not necessary to test multi-network capability
### Customization
The device plugin will register the SR-IOV enabled devices on the host, specified as
`rootDevices` in [sriov-conf.yaml](sriov-conf.yaml). Helper [systemd unit](systemd/sriov.service)
file is provided, which enables SR-IOV for the above `rootDevices`
> NOTE: This assumes homogenous nodes in the cluster
### Pre-req (SR-IOV only)
One each SR-IOV node make sure `VT-d` is enabled in the BIOS and `intel_iommu=on` on kernel commandline.
Setup systemd to bring up VFs on designated interfaces bound to network driver or `vfio-pci`
```bash
# Make sure vfio-pci is loaded on boot
echo 'vfio-pci' | sudo tee /etc/modules-load.d/sriov.conf
sudo systemctl restart systemd-modules-load.service
sudo cp systemd/sriov.sh /usr/bin/sriov.sh
sudo cp systemd/sriov.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now sriov.service
```
### Install
To install and configure `multus-cni` on all nodes, along with
`sriov-cni`, `vfioveth-cni` and `sriov-network-device-plugin`
```bash
kubectl apply -f .
kubectl get nodes -o json | jq '.items[].status.allocatable' # should list "intel.com/sriov_*"
```
## Tests
### Default only
To test if default connectivity is working
```bash
kubectl apply -f test/pod.yaml
kubectl exec test -- ip a # should see one interface only
```
### Bridge
To test multus with second interface created by `bridge` plugin
```bash
kubectl apply -f test/bridge
kubectl exec test-bridge -- ip a # should see two interfaces
ip a show mynet # bridge created on host if it doesnt exist already
```
### SR-IOV
To test multus with second interface created by `sriov` plugin
```bash
kubectl apply -f test/sriov
kubectl exec test-sriov -- ip a # second interface is a VF
kubectl exec test-sriov-dpdk -- ip a # veth pair with details of VF
kubectl exec test-sriov-dpdk -- ls -l /dev/vfio
```
+67
View File
@@ -0,0 +1,67 @@
#!/bin/bash -x
set -o errexit
set -o pipefail
set -o nounset
exec 3>&1
exec &>>/var/log/$(basename $0).log
PATH="$CNI_PATH:$(dirname "${BASH_SOURCE[0]}"):$PATH"
CNI_CONF=$(cat /dev/stdin)
get_peer_name() {
echo "$1-vdev"
}
get_mac_with_vfpci() {
local pf=$(readlink /sys/devices/pci*/*/$1/physfn | awk '{print substr($1,4)}')
local pfName=$(ls /sys/devices/pci*/*/$pf/net/ | head -1)
local idx=$(ls -l /sys/devices/pci*/*/$pf | awk -v vf=$1 'substr($11,4)==vf {print substr($9,7)}')
local mac=$(ip link show dev $pfName | awk -v idx="$idx" '$1=="vf" && $2==idx {print substr($4,1,17)}')
echo $mac
}
ipam() {
local plugin=$(echo $CNI_CONF | jq -r '.ipam.type')
local res=$(echo $"$CNI_CONF" | "$plugin" | jq -c '.')
echo $res
}
add_pair_ns() {
vfpci=$(echo $CNI_CONF | jq -r '.deviceID')
mac=$(get_mac_with_vfpci $vfpci)
peer=$(get_peer_name $CNI_IFNAME)
ip=$1
mkdir -p /var/run/netns/
ln -sfT $CNI_NETNS /var/run/netns/$CNI_CONTAINERID
ip netns exec $CNI_CONTAINERID ip link add $CNI_IFNAME type veth peer name $peer
ip netns exec $CNI_CONTAINERID ip link set $CNI_IFNAME addr $mac up
ip netns exec $CNI_CONTAINERID ip link set $peer up
ip netns exec $CNI_CONTAINERID ip addr add $ip dev $CNI_IFNAME
}
delete_pair_ns() {
ip netns exec $CNI_CONTAINERID ip link del $CNI_IFNAME
}
case $CNI_COMMAND in
ADD)
res=$(ipam)
ip=$(echo $res | jq -r '.ip4.ip')
add_pair_ns $ip
echo '{"cniVersion":"0.2.0"}' | jq -c --arg ip $ip '.ip4.ip = $ip' >&3
;;
DEL)
set +o errexit
ipam
delete_pair_ns
set -o errexit
;;
*)
echo "CNI_COMMAND=[ADD|DEL] only supported"
exit 1
;;
esac
@@ -0,0 +1,212 @@
---
apiVersion: apiextensions.k8s.io/v1beta1
kind: CustomResourceDefinition
metadata:
# name must match the spec fields below, and be in the form: <plural>.<group>
name: network-attachment-definitions.k8s.cni.cncf.io
spec:
# group name to use for REST API: /apis/<group>/<version>
group: k8s.cni.cncf.io
# version name to use for REST API: /apis/<group>/<version>
version: v1
# either Namespaced or Cluster
scope: Namespaced
names:
# plural name to be used in the URL: /apis/<group>/<version>/<plural>
plural: network-attachment-definitions
# singular name to be used as an alias on the CLI and for display
singular: network-attachment-definition
# kind is normally the CamelCased singular type. Your resource manifests use this.
kind: NetworkAttachmentDefinition
# shortNames allow shorter string to match your resource on the CLI
shortNames:
- net-attach-def
---
apiVersion: v1
kind: ServiceAccount
metadata:
name: multus-sa
namespace: kube-system
---
apiVersion: v1
kind: Secret
metadata:
name: multus-sa-secret
namespace: kube-system
annotations:
kubernetes.io/service-account.name: multus-sa
type: kubernetes.io/service-account-token
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: multus-pod-networks-lister
rules:
- apiGroups: [""]
resources: ["pods"]
verbs: ["get"]
- apiGroups: [""]
resources: ["pods/status"]
verbs: ["update"]
- apiGroups: ["k8s.cni.cncf.io"]
resources: ["*"]
verbs: ["get"]
---
kind: ConfigMap
apiVersion: v1
metadata:
name: multus-scripts
namespace: kube-system
data:
install-certs.sh: |
# Copied from Calico
# https://github.com/projectcalico/cni-plugin/blob/master/k8s-install/scripts/install-cni.sh
touch /host/etc/cni/net.d/multus-kubeconfig
chmod 600 /host/etc/cni/net.d/multus-kubeconfig
SERVICE_ACCOUNT_PATH=/var/run/secrets/multus/serviceaccount
KUBE_CA_FILE=$SERVICE_ACCOUNT_PATH/ca.crt
TLS_CFG="certificate-authority-data: $(cat $KUBE_CA_FILE | base64 | tr -d '\n')"
SERVICEACCOUNT_TOKEN=$(cat $SERVICE_ACCOUNT_PATH/token)
cat > /host/etc/cni/net.d/multus-kubeconfig <<EOF
# Kubeconfig file for Multus CNI plugin.
apiVersion: v1
kind: Config
clusters:
- name: local
cluster:
server: https://${KUBERNETES_SERVICE_HOST}:${KUBERNETES_SERVICE_PORT}
$TLS_CFG
users:
- name: multus
user:
token: "${SERVICEACCOUNT_TOKEN}"
contexts:
- name: multus-context
context:
cluster: local
user: multus
current-context: multus-context
EOF
install-multus-conf.sh: |
# copied from https://github.com/intel/multus-cni/blob/master/images/entrypoint.sh
rm -f /host/etc/cni/net.d/00-multus.conf
MASTER_PLUGIN="$(ls /host/etc/cni/net.d | grep -E '\.conf(list)?$' | head -1)"
MASTER_PLUGIN_JSON="$(cat /host/etc/cni/net.d/$MASTER_PLUGIN)"
cat > /host/etc/cni/net.d/00-multus.conf <<EOF
{
"name": "multus-cni-network",
"type": "multus",
"logFile": "/var/log/multus.log",
"logLevel": "debug",
"kubeconfig": "/etc/cni/net.d/multus-kubeconfig",
"delegates": [
$MASTER_PLUGIN_JSON
]
}
EOF
---
kind: ClusterRoleBinding
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: multus-rb
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: multus-pod-networks-lister
subjects:
- kind: ServiceAccount
name: multus-sa
namespace: kube-system
---
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: sriov-device-plugin
namespace: kube-system
spec:
selector:
matchLabels:
name: sriov-device-plugin
template:
metadata:
labels:
name: sriov-device-plugin
spec:
initContainers:
- name: multus
image: krsna1729/multus-sriov:k8s-1.13
command: [ "bash", "-c" ]
args:
- cp /tmp/cni/bin/{multus,sriov,vfioveth,jq} /host/opt/cni/bin/;
/tmp/multus/install-multus-conf.sh;
/tmp/multus/install-certs.sh;
echo "Restarting crio kubelet";
systemctl restart crio; # Needed when crio manages ns lifecycle
systemctl restart kubelet;
volumeMounts:
- name: usr-bin
mountPath: /host/usr/bin
- name: cni-bin
mountPath: /host/opt/cni/bin
- name: multus-sa
mountPath: /var/run/secrets/multus/serviceaccount
- name: multus-scripts
mountPath: /tmp/multus
- name: cni-conf
mountPath: /host/etc/cni/net.d
- name: dbus
mountPath: /var/run/dbus
- name: systemd
mountPath: /run/systemd
containers:
- name: sriovdp
image: krsna1729/multus-sriov:k8s-1.13
command: [ "sh", "-c" ]
args:
- /usr/bin/sriovdp --logtostderr -v 10;
sleep infinity;
volumeMounts:
- name: net
mountPath: /sys/class/net
readOnly: true
- name: dp-sock
mountPath: /var/lib/kubelet/device-plugins/
readOnly: false
- mountPath: /etc/pcidp
name: sriov-config
readOnly: false
volumes:
- name: sriov-config
configMap:
name: sriov-config
- name: usr-bin
hostPath:
path: /usr/bin
- name: cni-bin
hostPath:
path: /opt/cni/bin
- name: multus-sa
secret:
secretName: multus-sa-secret
- name: multus-scripts
configMap:
defaultMode: 511
name: multus-scripts
- name: cni-conf
hostPath:
path: /etc/cni/net.d
- name: dbus
hostPath:
path: /var/run/dbus
- name: systemd
hostPath:
path: /run/systemd
- name: net
hostPath:
path: /sys/class/net
- name: dp-sock
hostPath:
path: /var/lib/kubelet/device-plugins/
hostNetwork: true
hostPID: true
@@ -0,0 +1,25 @@
---
kind: ConfigMap
apiVersion: v1
metadata:
name: sriov-config
namespace: kube-system
data:
config.json: |
{
"resourceList":
[
{
"resourceName": "sriov_netdevice",
"rootDevices": ["07:00.0"],
"sriovMode": true,
"deviceType": "netdevice"
},
{
"resourceName": "sriov_vfio",
"rootDevices": ["07:00.1"],
"sriovMode": true,
"deviceType": "vfio"
}
]
}
@@ -0,0 +1,10 @@
[Unit]
Description=Create VFs on ens785f0 (netdev) ens785f1 (vfio) interfaces
[Service]
Type=oneshot
ExecStart=/usr/bin/sriov.sh ens785f0
ExecStart=/usr/bin/sriov.sh -b ens785f1
[Install]
WantedBy=default.target
+58
View File
@@ -0,0 +1,58 @@
#!/bin/bash
set -o errexit
set -o pipefail
set -o nounset
OPTIND=1
bind="false"
while getopts ":b" opt; do
case ${opt} in
b)
bind="true"
;;
\?)
echo "Usage: sriov.sh [-b] ens785f0 ens785f1 ..."
echo "-b Bind to vfio-pci"
exit
;;
esac
done
shift $((OPTIND - 1))
setup_pf() {
local pf=$1
echo "Resetting PF $pf"
echo 0 | tee /sys/class/net/$pf/device/sriov_numvfs
local NUM_VFS=$(cat /sys/class/net/$pf/device/sriov_totalvfs)
echo "Enabling $NUM_VFS VFs for $pf"
echo $NUM_VFS | tee /sys/class/net/$pf/device/sriov_numvfs
ip link set $pf up
sleep 1
}
setup_vfs() {
local pf=$1
local pfpci=$(readlink /sys/devices/pci*/*/*/net/$pf/device | awk '{print substr($1,10)}')
local NUM_VFS=$(cat /sys/class/net/$pf/device/sriov_numvfs)
for ((idx = 0; idx < NUM_VFS; idx++)); do
ip link set dev $pf vf $idx state enable
if [ $bind != "true" ]; then continue; fi
local vfn="virtfn$idx"
local vfpci=$(ls -l /sys/devices/pci*/*/$pfpci | awk -v vfn=$vfn 'vfn==$9 {print substr($11,4)}')
# Capture and set MAC of the VF before unbinding from linux, for later use in CNI
local mac=$(cat /sys/bus/pci*/*/$vfpci/net/*/address)
ip link set dev $pf vf $idx mac $mac
# Bind VF to vfio-pci
echo $vfpci >/sys/bus/pci*/*/$vfpci/driver/unbind
echo "vfio-pci" >/sys/devices/pci*/*/$vfpci/driver_override
echo $vfpci >/sys/bus/pci/drivers/vfio-pci/bind
done
}
for pf in "$@"; do
setup_pf $pf
setup_vfs $pf
done
@@ -0,0 +1,16 @@
---
apiVersion: "k8s.cni.cncf.io/v1"
kind: NetworkAttachmentDefinition
metadata:
name: mynet
spec:
config: '{
"name": "mynet",
"type": "bridge",
"bridge": "mynet",
"ipam": {
"type": "host-local",
"subnet": "198.18.0.0/24"
}
}'
@@ -0,0 +1,14 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test-bridge
annotations:
k8s.v1.cni.cncf.io/networks: '[
{ "name": "mynet", "interface": "mynet" }
]'
spec:
containers:
- name: busy
image: busybox
command: [ "top" ]
@@ -0,0 +1,10 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test
spec:
containers:
- name: busy
image: busybox
command: [ "top" ]
@@ -0,0 +1,39 @@
---
apiVersion: "k8s.cni.cncf.io/v1"
kind: NetworkAttachmentDefinition
metadata:
name: sriov-net
annotations:
k8s.v1.cni.cncf.io/resourceName: intel.com/sriov_netdevice
spec:
config: '{
"type": "sriov",
"name": "sriov-net",
"ipam": {
"type": "host-local",
"subnet": "198.19.0.0/24",
"rangeStart": "198.19.0.100",
"rangeEnd": "198.19.0.200",
"gateway": "198.19.0.1"
}
}'
---
apiVersion: "k8s.cni.cncf.io/v1"
kind: NetworkAttachmentDefinition
metadata:
name: sriov-net-dpdk
annotations:
k8s.v1.cni.cncf.io/resourceName: intel.com/sriov_vfio
spec:
config: '{
"type": "vfioveth",
"name": "sriov-net",
"ipam": {
"type": "host-local",
"subnet": "198.19.0.0/24",
"rangeStart": "198.19.0.100",
"rangeEnd": "198.19.0.200",
"gateway": "198.19.0.1"
}
}'
@@ -0,0 +1,31 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test-sriov
annotations:
k8s.v1.cni.cncf.io/networks: sriov-net
spec:
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
limits:
intel.com/sriov_netdevice: '1'
---
apiVersion: v1
kind: Pod
metadata:
name: test-sriov-dpdk
annotations:
k8s.v1.cni.cncf.io/networks: sriov-net-dpdk
spec:
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
limits:
intel.com/sriov_vfio: '1'
+27 -49
View File
@@ -1,13 +1,24 @@
# How to setup the cluster
## Prerequisite
This setup currently will work with k8s 1.14 & above. Any version of k8s before that might work, but is not guaranteed.
## Sample multi-node vagrant setup
To be able to test this tool, you can create a 3-node vagrant setup. In this tutorial, we will talk about using libvirt, but you can use any hypervisor that you are familiar with.
To be able to test this tool, you can create a 3-node vagrant setup. In this tutorial, we will talk about using [libvirt](https://github.com/vagrant-libvirt/vagrant-libvirt), but you can use any hypervisor that you are familiar with.
* Install vagrant on the distro you are using. Steps can be found at [Vagrant docs](https://www.vagrantup.com/intro/getting-started/install.html#installing-vagrant)
* `vagrant up --provider=libvirt`
Now you have a 3 node cluster up and running. Each of them have 2 vCPU, 4GB Memory, 2x10GB disks, 1 additional private network.
Customize the setup using environment variables. E.g., `NODES=1 MEMORY=8192 CPUS=8 vagrant up --provider=libvirt`
To login to the master node and change to this directory
```bash
vagrant ssh clr-01
cd clr-k8s-examples
```
## Setup the nodes in the cluster
@@ -20,18 +31,17 @@ This script ensures the following
* Customizes the system to ensure correct defaults are setup (IP Forwarding, Swap off,...)
* Ensures all the dependencies are loaded on boot (kernel modules)
> NOTE: This step is done automatically if using vagrant.
> NOTE: This step is done automatically if using vagrant.
### Enabling experimental firecracker support
> EXPERIMENTAL: Optionally run [`setup_firecracker.sh`](setup_firecracker.sh) to be
> EXPERIMENTAL: Optionally run [`setup_kata_firecracker.sh`](setup_kata_firecracker.sh) to be
able to use firecracker VMM with Kata.
The firecracker setup switches the setup to use a sparse file backed loop device for
devicemapper storage. This should not be used for production.
> NOTE: This step is done automatically if using vagrant.
> NOTE: This step is done automatically if using vagrant.
## Bring up the master
@@ -40,13 +50,12 @@ master and also uses kubelet config via [`kubeadm.yaml`](kubeadm.yaml)
to propagate cluster wide kubelet configuration to all workers. Customize it if
you need to setup other cluster wide properties.
There are two flavors of install -
- `minimal`: initialize cluster, add kata runtimeclass, install canal CNI and metrics server
- `all`: minimal, install rook storage, prometheus, ELK, nginx-ingress, etc.,
There are different flavors to install, run `./create_stack.sh help` to get
more information.
```bash
# default is 'all'
./create_stack.sh [minimal|all]
# default shows help
./create_stack.sh <subcommand>
```
## Join Workers to the cluster
@@ -73,54 +82,23 @@ runtime class set to "kata" will launch the POD/Deployment with Kata.
An example is
`kubectl apply -f tests/test-deploy-kata.yaml`
`kubectl apply -f tests/deploy-svc-ing/test-deploy-kata-qemu.yaml`
### Running Kata Workloads with Firecracker
> EXPERIMENTAL: If firecracker setup has been enabled, runtime class set to "fire" will launch the POD/Deployment
> EXPERIMENTAL: If firecracker setup has been enabled, runtime class set to "kata-fc" will launch the POD/Deployment
with firecracker as the isolation mechanism for Kata.
An example is
`kubectl apply -f tests/test-deploy-fire.yaml`
`kubectl apply -f tests/deploy-svc-ing/test-deploy-kata-fc.yaml`
## Making Kata the default runtime
## Making Kata the default runtime using admission controller
Today in `crio.conf` runc is the default runtime when a user does not specify
`runtimeClass` in the pod spec. If you want to run a cluster where kata is used
by default, except for workloads we know for sure will not work with kata, use
the [admission webhook](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks)
and sample admission controller we created by running -
`kubectl apply -f admit-kata/`
The [admission webhook](admit-kata/webhook-registration.yaml)
is setup to exclude certian namespaces from being run with Kata using filters on namespace labels.
```yaml
namespaceSelector:
matchExpressions:
- {key: "kata", operator: NotIn, values: ["false"]}
```
The rook operators for example are marked as such
```yaml
apiVersion: v1
kind: Namespace
metadata:
name: rook-ceph-system
labels:
kata: "false"
```
Pods not explicitly excluded by the namespace filter are dynamically tagged to
run with Kata with some [exceptions](https://github.com/mcastelino/kubewebhook/blob/topic/hack-kata/examples/pod-annotate/main.go#L25) -
* `hostNetwork: true`
* `rook-ceph` and `rook-ceph-system` namespaces (buggy)
Other pod properties will be added as exceptions in future.
If you want to run a cluster where kata is used
by default, except for workloads we know for sure will not work with kata, using
[admission webhook](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks)
and sample admission controller, follow `admit-kata` [README.md](admit-kata/README.md)
## Accessing control plane services
+30 -21
View File
@@ -4,21 +4,22 @@
require 'fileutils'
require 'ipaddr'
require 'securerandom'
DISK_UUID = SecureRandom.urlsafe_base64(9)
$num_instances = (ENV['NODES'] || 3).to_i
$cpus = (ENV['CPUS'] || 2).to_i
$memory = (ENV['MEMORY'] || 4096).to_i
$disks = 2
# Using folder prefix instead of uuid until vagrant-libvirt fixes disk cleanup
$disk_prefix = File.basename(File.dirname(__FILE__), "/")
$disk_size = "10G"
$box = "gmmaha/clearlinux"
$loader = File.join(File.dirname(__FILE__), "OVMF.fd")
$box = "AntonioMeireles/ClearLinux"
File.exists?("/usr/share/qemu/OVMF.fd") ? $loader = "/usr/share/qemu/OVMF.fd" : $loader = File.join(File.dirname(__FILE__), "OVMF.fd")
$vm_name_prefix = "clr"
base_ip = IPAddr.new("192.52.100.10")
hosts = {}
proxy_ip_list = ""
#DISK_UUID = Time.now.utc.to_i
driveletters = ('a'..'z').to_a
$base_ip = IPAddr.new("192.52.100.10")
$hosts = {}
$proxy_ip_list = ""
$driveletters = ('a'..'z').to_a
$setup_fc = true ? (['true', '1'].include? ENV['SETUP_FC'].to_s) : false
if not File.exists?($loader)
system('curl -O https://download.clearlinux.org/image/OVMF.fd')
@@ -43,39 +44,47 @@ Vagrant.configure("2") do |config|
# boxes at https://vagrantcloud.com/search.
config.vm.box = $box
# Mount the current dir at home folder instead of default
config.vm.synced_folder './', '/vagrant', disabled: true
config.vm.synced_folder './', '/home/clear/' + File.basename(Dir.getwd), type: 'rsync',
rsync__args: ["--verbose", "--archive", "--delete", "-zz", "--copy-links"]
#Setup proxies for all machines
(1..$num_instances).each do |i|
base_ip = base_ip.succ
hosts["clr-%02d" % i] = base_ip.to_s
$base_ip = $base_ip.succ
$hosts["clr-%02d" % i] = $base_ip.to_s
end
hosts.each do |vm_name, ip|
$hosts.each do |vm_name, ip|
proxy_ip_list = ("#{proxy_ip_list},#{vm_name},#{ip}")
end
hosts.each do |vm_name, ip|
$hosts.each do |vm_name, ip|
config.vm.define vm_name do |c|
c.vm.hostname = vm_name
c.vm.network :private_network, ip: ip, autostart: true
c.vm.provider :libvirt do |lv|
lv.loader = $loader
lv.cpu_mode = "host-passthrough"
lv.nested = true
lv.loader = $loader
lv.cpus = $cpus
lv.memory = $memory
(1..$disks).each do |d|
lv.storage :file, :device => "hd#{driveletters[d]}", :path => "disk-#{vm_name}-#{d}-#{DISK_UUID}.disk", :size => $disk_size, :type => "raw"
lv.storage :file, :device => "hd#{$driveletters[d]}", :path => "disk-#{$disk_prefix}-#{vm_name}-#{d}.disk", :size => $disk_size, :type => "raw"
end
end
if Vagrant.has_plugin?("vagrant-proxyconf")
c.proxy.http = (ENV['http_proxy']||ENV['HTTP_PROXY'])
c.proxy.https = (ENV['https_proxy']||ENV['HTTPS_PROXY'])
c.proxy.no_proxy = (ENV['no_proxy']+"#{proxy_ip_list}" || ENV['NO_PROXY']+"#{proxy_ip_list}" || "localhost,127.0.0.1,172.16.10.10#{proxy_ip_list}")
if ENV['http_proxy'] || ENV['HTTP_PROXY']
if Vagrant.has_plugin?("vagrant-proxyconf")
c.proxy.http = (ENV['http_proxy']||ENV['HTTP_PROXY'])
c.proxy.https = (ENV['https_proxy']||ENV['HTTPS_PROXY'])
c.proxy.no_proxy = (ENV['no_proxy']+"#{proxy_ip_list}" || ENV['NO_PROXY']+"#{proxy_ip_list}" || "localhost,127.0.0.1,172.16.10.10#{proxy_ip_list}")
end
end
# Bad hack for the vagrant libvirt boxes. WIll be removed once they are fixed.
c.vm.provision "shell", privileged: false, inline: "sudo usermod --password vagrant root"
c.vm.provision "shell", privileged: false, path: "setup_system.sh"
c.vm.provision "shell", privileged: false, path: "setup_firecracker.sh"
if $setup_fc
c.vm.provision "shell", privileged: false, path: "setup_kata_firecracker.sh"
end
# Include shells bundle to get bash completion and add kubectl's commands to vagrant's shell
c.vm.provision "shell", privileged: false, inline: 'sudo -E swupd bundle-add shells; echo "source <(kubectl completion bash)" >> $HOME/.bashrc'
end
end
end
+39
View File
@@ -0,0 +1,39 @@
# Kata Admission controller webhook
Implement a simple admission controller webhook to annotate pods with the
Kata runtime class.
## How to build the admission controller
First build the admission controller image and the associated
Kubernetes yaml files required to instantiate the admission
controller.
```bash
$ docker build -t katadocker/kata-webhook-example:latest .
$ ./create_certs.sh
```
> **Note:**
> Image needs to be published for the webhook needs to work. Alternately
> on a single machine cluster change the `imagePullPolicy` to use the locally
> built image.
## Making Kata the default runtime using an admission controller
Today in `crio.conf` `runc` is the default runtime when a user does not specify
`runtimeClass` in the pod spec. If you want to run a cluster where Kata is used
by default, except for workloads we know for sure will not work with Kata, use
the [admission webhook](https://kubernetes.io/docs/reference/access-authn-authz/extensible-admission-controllers/#admission-webhooks)
and sample admission controller we created by running
```bash
$ kubectl apply -f deploy/
```
The webhook mutates pods to use the kata runtime class for all pods except
those with
* `hostNetwork: true`
* namespace: `rook-ceph` and `rook-ceph-system`
+28
View File
@@ -0,0 +1,28 @@
#! /bin/bash
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
WEBHOOK_NS=${1:-"default"}
WEBHOOK_NAME=${2:-"pod-annotate"}
WEBHOOK_SVC="${WEBHOOK_NAME}-webhook"
# Create certs for our webhook
openssl genrsa -out webhookCA.key 2048
openssl req -new -key ./webhookCA.key -subj "/CN=${WEBHOOK_SVC}.${WEBHOOK_NS}.svc" -out ./webhookCA.csr
openssl x509 -req -days 365 -in webhookCA.csr -signkey webhookCA.key -out webhook.crt
# Create certs secrets for k8s
kubectl create secret generic \
${WEBHOOK_SVC}-certs \
--from-file=key.pem=./webhookCA.key \
--from-file=cert.pem=./webhook.crt \
--dry-run -o yaml > ./deploy/webhook-certs.yaml
# Set the CABundle on the webhook registration
CA_BUNDLE=$(cat ./webhook.crt | base64 -w0)
sed "s/CA_BUNDLE/${CA_BUNDLE}/" ./deploy/webhook-registration.yaml.tpl > ./deploy/webhook-registration.yaml
# Clean
rm ./webhookCA* && rm ./webhook.crt
@@ -0,0 +1,24 @@
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
apiVersion: admissionregistration.k8s.io/v1beta1
kind: MutatingWebhookConfiguration
metadata:
name: pod-annotate-webhook
labels:
app: pod-annotate-webhook
kind: mutator
webhooks:
- name: pod-annotate-webhook.kata.xyz
clientConfig:
service:
name: pod-annotate-webhook
namespace: default
path: "/mutate"
caBundle: CA_BUNDLE
rules:
- operations: [ "CREATE" ]
apiGroups: [""]
apiVersions: ["v1"]
resources: ["pods"]
@@ -1,3 +1,7 @@
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
@@ -13,11 +17,12 @@ spec:
spec:
containers:
- name: pod-annotate-webhook
image: mcastelino/kubewebhook-pod-annotate-example:1.0
image: katadocker/kata-webhook-example:latest
imagePullPolicy: Always
args:
- -tls-cert-file=/etc/webhook/certs/cert.pem
- -tls-key-file=/etc/webhook/certs/key.pem
- -exclude-namespaces=rook-ceph-system,rook-ceph
volumeMounts:
- name: webhook-certs
mountPath: /etc/webhook/certs
+2
View File
@@ -0,0 +1,2 @@
https://github.com/kata-containers/tests/tree/master/kata-webhook
Commit: 5ad2cec
@@ -1,8 +0,0 @@
apiVersion: v1
data:
cert.pem: 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
key.pem: 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
kind: Secret
metadata:
creationTimestamp: null
name: pod-annotate-webhook-certs
@@ -1,24 +0,0 @@
apiVersion: admissionregistration.k8s.io/v1beta1
kind: MutatingWebhookConfiguration
metadata:
name: pod-annotate-webhook
labels:
app: pod-annotate-webhook
kind: mutator
webhooks:
- name: pod-annotate-webhook.pod.xyz
clientConfig:
service:
name: pod-annotate-webhook
namespace: default
path: "/mutate"
caBundle: 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
rules:
- operations: [ "CREATE" ]
apiGroups: [""]
apiVersions: ["v1"]
resources: ["pods"]
namespaceSelector:
matchExpressions:
- {key: "kata", operator: NotIn, values: ["false"]}
+46 -33
View File
@@ -8,12 +8,23 @@ CUR_DIR=$(pwd)
SCRIPT_DIR="$(dirname "${BASH_SOURCE[0]}")"
function print_usage_exit() {
echo $"Usage: $0 [minimal|all]"
exit 1
exit_code=${1:-0}
cat <<EOT
Usage: $0 [subcommand]
Subcommands:
$(
for cmd in "${!command_handlers[@]}"; do
printf "\t%s:|\t%s\n" "${cmd}" "${command_help[${cmd}]:-Not-documented}"
done | sort | column -t -s "|"
)
EOT
exit "${exit_code}"
}
function finish() {
cd $CUR_DIR
cd "${CUR_DIR}"
}
trap finish EXIT
@@ -22,29 +33,26 @@ function cluster_init() {
#to enable the RuntimeClass featuregate
sudo -E kubeadm init --config=./kubeadm.yaml
rm -rf "${HOME}/.kube"
mkdir -p "${HOME}/.kube"
sudo cp -i /etc/kubernetes/admin.conf "${HOME}/.kube/config"
sudo chown "$(id -u):$(id -g)" "${HOME}/.kube/config"
# If this an interactive terminal then wait for user to join workers
if [ -t 0 ]; then
read -p "Join other nodes. Press enter to continue"
fi
rm -rf $HOME/.kube
mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config
#Ensure single node k8s works
if [ $(kubectl get nodes | wc -l) -eq 2 ]; then
if [ "$(kubectl get nodes | wc -l)" -eq 2 ]; then
kubectl taint nodes --all node-role.kubernetes.io/master-
fi
}
function runtimeclass_kata() {
#Add support for kata runtime
kubectl apply -f 8-kata/runtimeclass_crd.yaml
while [[ $(kubectl get crd runtimeclasses.node.k8s.io >/dev/null 2>&1) || $? -ne 0 ]]; do
echo "Waiting for runtime class CRD"
sleep 2
done
function kata() {
# Install kata artifacts using kata-deploy
kubectl apply -f 8-kata/deploy/kata-rbac.yaml
kubectl apply -f 8-kata/deploy/kata-deploy.yaml
kubectl apply -f 8-kata/
}
@@ -100,8 +108,8 @@ function miscellaneous() {
function minimal() {
cluster_init
runtimeclass_kata
cni
kata
metrics
}
@@ -112,20 +120,25 @@ function all() {
miscellaneous
}
cd $SCRIPT_DIR
if [[ "$#" -eq 0 ]]; then
all
exit
fi
declare -A command_handlers
command_handlers[init]=cluster_init
command_handlers[cni]=cni
command_handlers[minimal]=minimal
command_handlers[all]=all
command_handlers[help]=print_usage_exit
case "$1" in
minimal)
minimal
;;
all)
all
;;
*)
print_usage_exit
;;
esac
declare -A command_help
command_help[init]="Only inits a cluster using kubeadm"
command_help[cni]="Setup network for running cluster"
command_help[minimal]="init + cni + kata + metrics"
command_help[all]="minimal + storage + monitoring + miscellaneous"
command_help[help]="show this message"
cd "${SCRIPT_DIR}"
cmd_handler=${command_handlers[${1:-none}]:-unimplemented}
if [ "${cmd_handler}" != "unimplemented" ]; then
"${cmd_handler}"
else
print_usage_exit 1
fi
+4 -11
View File
@@ -1,15 +1,10 @@
apiVersion: kubeadm.k8s.io/v1alpha3
apiVersion: kubeadm.k8s.io/v1beta1
kind: InitConfiguration
nodeRegistration:
criSocket: /var/run/crio/crio.sock
---
apiVersion: kubelet.config.k8s.io/v1beta1
kind: KubeletConfiguration
featureGates:
RuntimeClass: true
# Kata does not work with static
# https://github.com/kata-containers/runtime/issues/878
# cpuManagerPolicy: static
# Allowing for CPU pinning and isolation in case of guaranteed QoS class
cpuManagerPolicy: static
systemReserved:
cpu: 500m
memory: 256M
@@ -17,11 +12,9 @@ kubeReserved:
cpu: 500m
memory: 256M
---
apiVersion: kubeadm.k8s.io/v1alpha3
apiVersion: kubeadm.k8s.io/v1beta1
kind: ClusterConfiguration
networking:
dnsDomain: cluster.local
podSubnet: 10.244.0.0/16
serviceSubnet: 10.96.0.0/12
apiServerExtraArgs:
feature-gates: RuntimeClass=true
+13 -6
View File
@@ -3,7 +3,10 @@
set -o nounset
#Cleanup
sudo -E kubeadm reset -f --cri-socket="/var/run/crio/crio.sock"
reset_cluster() {
sudo -E kubeadm reset -f
}
reset_cluster
for ctr in $(sudo crictl ps --quiet); do
sudo crictl stop "$ctr"
@@ -15,9 +18,10 @@ for pod in $(sudo crictl pods --quiet); do
done
#Forcefull cleanup all artifacts
#This is needed is things really go wrong
#This is needed if things really go wrong
sudo systemctl stop kubelet
sudo systemctl stop crio
systemctl is-active crio && sudo systemctl stop crio
systemctl is-active containerd && sudo systemctl stop containerd
sudo pkill -9 qemu
sudo pkill -9 kata
sudo pkill -9 kube
@@ -39,8 +43,11 @@ sudo -E bash -c "rm -r /var/run/kata-containers/*"
sudo rm -rf /var/lib/rook
sudo systemctl daemon-reload
sudo systemctl enable kubelet crio
sudo systemctl restart crio
sudo systemctl is-active crio && sudo systemctl stop crio
sudo systemctl is-active containerd && sudo systemctl stop containerd
sudo systemctl is-enabled crio && sudo systemctl restart crio
sudo systemctl is-enabled containerd && sudo systemctl restart containerd
sudo systemctl restart kubelet
sudo -E kubeadm reset -f --cri-socket="/var/run/crio/crio.sock"
reset_cluster
-92
View File
@@ -1,92 +0,0 @@
#!/bin/bash
set -o errexit
set -o pipefail
set -o nounset
sudo mkdir -p /etc/kata-containers
# Setup a configuration to be used by firecracker
cat <<EOT | sudo tee /etc/kata-containers/configuration_firecracker.toml
[hypervisor.firecracker]
path = "/usr/bin/firecracker"
kernel = "/usr//share/kata-containers/vmlinux.container"
image = "/usr//share/kata-containers/kata-containers.img"
kernel_params = ""
default_vcpus = 1
default_memory = 4096
default_maxvcpus = 0
default_bridges = 1
block_device_driver = "virtio-mmio"
disable_block_device_use = false
enable_debug = true
use_vsock = true
[shim.kata]
path = "/usr//libexec/kata-containers/kata-shim"
[agent.kata]
[runtime]
internetworking_model="tcfilter"
EOT
# Firecracker can only work with devicemapper
# Setup a sparse disk to be used for devicemapper
sudo rm -f /var/lib/crio/devicemapper/disk.img
sudo mkdir -p /var/lib/crio/devicemapper
sudo truncate /var/lib/crio/devicemapper/disk.img --size 10G
# Ensure that this disk is loop mounted at each boot
sudo mkdir -p /etc/systemd/system
cat <<EOT | sudo tee /etc/systemd/system/devicemapper.service
[Unit]
Description=Setup CRIO devicemapper
DefaultDependencies=no
After=systemd-udev-settle.service
Before=lvm2-activation-early.service
Wants=systemd-udev-settle.service
[Service]
ExecStart=-/sbin/losetup /dev/loop8 /var/lib/crio/devicemapper/disk.img
RemainAfterExit=true
Type=oneshot
[Install]
WantedBy=local-fs.target
EOT
sudo systemctl daemon-reload
sudo systemctl enable --now devicemapper
# For now till we address https://github.com/kubernetes-sigs/cri-o/issues/1991
# use a shell script to expose firecracker through kata
cat <<EOT | sudo tee /usr/bin/kata-runtime-fire
#!/bin/bash
/usr/bin/kata-runtime --kata-config /etc/kata-containers/configuration_firecracker.toml "\$@"
EOT
sudo chmod +x /usr/bin/kata-runtime-fire
# Add firecracker as a second runtime
# Also setup crio to use devicemapper
sudo mkdir -p /etc/crio/
sudo cp /usr/share/defaults/crio/crio.conf /etc/crio/crio.conf
echo -e "\n[crio.runtime.runtimes.kata]\nruntime_path = \"/usr/bin/kata-runtime\"" | sudo tee -a /etc/crio/crio.conf
echo -e "\n[crio.runtime.runtimes.fire]\nruntime_path = \"/usr/bin/kata-runtime-fire\"" | sudo tee -a /etc/crio/crio.conf
sudo sed -i 's|\(\[crio\.runtime\]\)|\1\nmanage_network_ns_lifecycle = true|' /etc/crio/crio.conf
sudo sed -i 's/storage_driver = \"overlay\"/storage_driver = \"devicemapper\"\
storage_option = [\
\"dm.basesize=8G\",\
\"dm.directlvm_device=\/dev\/loop8\",\
\"dm.directlvm_device_force=true\",\
\"dm.fs=ext4\"\
]/g' /etc/crio/crio.conf
sudo systemctl restart crio || true
+45
View File
@@ -0,0 +1,45 @@
#!/bin/bash
set -o errexit
set -o pipefail
set -o nounset
# Firecracker can only work with devicemapper
# Setup a sparse disk to be used for devicemapper
sudo rm -f /var/lib/crio/devicemapper/disk.img
sudo mkdir -p /var/lib/crio/devicemapper
sudo truncate /var/lib/crio/devicemapper/disk.img --size 10G
# Ensure that this disk is loop mounted at each boot
sudo mkdir -p /etc/systemd/system
cat <<EOT | sudo tee /etc/systemd/system/devicemapper.service
[Unit]
Description=Setup CRIO devicemapper
DefaultDependencies=no
After=systemd-udev-settle.service
Before=lvm2-activation-early.service
Wants=systemd-udev-settle.service
[Service]
ExecStart=-/sbin/losetup /dev/loop8 /var/lib/crio/devicemapper/disk.img
RemainAfterExit=true
Type=oneshot
[Install]
WantedBy=local-fs.target
EOT
sudo systemctl daemon-reload
sudo systemctl enable --now devicemapper
sudo sed -i 's/storage_driver = \"overlay\"/storage_driver = \"devicemapper\"\
storage_option = [\
\"dm.basesize=8G\",\
\"dm.directlvm_device=\/dev\/loop8\",\
\"dm.directlvm_device_force=true\",\
\"dm.override_udev_sync_check=true",\
\"dm.fs=ext4\"\
]/g' /etc/crio/crio.conf
sudo systemctl restart crio || true
+6 -2
View File
@@ -22,6 +22,8 @@ fi
sudo mkdir -p /etc/sysctl.d/
cat <<EOT | sudo bash -c "cat > /etc/sysctl.d/60-k8s.conf"
net.ipv4.ip_forward=1
net.ipv4.conf.default.rp_filter=1
net.ipv4.conf.all.rp_filter=1
EOT
sudo systemctl restart systemd-sysctl
@@ -33,6 +35,10 @@ vhost_vsock
overlay
EOT
# Make sure /etc/hosts file exists
if [ ! -f /etc/hosts ]; then
sudo touch /etc/hosts
fi
hostcount=$(grep '127.0.0.1 localhost' /etc/hosts | wc -l)
if [ "$hostcount" == "0" ]; then
echo "127.0.0.1 localhost $(hostname)" | sudo bash -c "cat >> /etc/hosts"
@@ -46,8 +52,6 @@ sudo systemctl daemon-reload
echo "The following kubelet command may complain... it is not an error"
sudo systemctl enable --now kubelet crio || true
sudo mkdir -p /usr/libexec/cni /opt/cni
[ ! -e /opt/cni/bin/cni ] && sudo ln -s /usr/libexec/cni /opt/cni/bin
#Ensure that the system is ready without requiring a reboot
sudo swapoff -a
sudo systemctl restart systemd-modules-load.service
@@ -0,0 +1,12 @@
apiVersion: v1
kind: Pod
metadata:
name: load-generator
spec:
containers:
- command: ["/bin/sh", "-c"]
args:
- while true; do wget -q -O- http://php-apache-test; done;
image: busybox
imagePullPolicy: Always
name: load-generator
+54
View File
@@ -0,0 +1,54 @@
---
apiVersion: apps/v1
kind: Deployment
metadata:
labels:
run: php-apache-test
name: php-apache-test
spec:
selector:
matchLabels:
run: php-apache-test
template:
metadata:
labels:
run: php-apache-test
spec:
containers:
- image: k8s.gcr.io/hpa-example
name: php-apache-test
ports:
- containerPort: 80
protocol: TCP
resources:
requests:
cpu: 200m
restartPolicy: Always
---
apiVersion: v1
kind: Service
metadata:
name: php-apache-test
spec:
ports:
- port: 80
protocol: TCP
targetPort: 80
selector:
run: php-apache-test
sessionAffinity: None
type: ClusterIP
---
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
name: php-apache-test
spec:
maxReplicas: 10
minReplicas: 1
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: php-apache-test
targetCPUUtilizationPercentage: 50
+5
View File
@@ -0,0 +1,5 @@
#!/bin/bash
SCRIPT_DIR="$(dirname "${BASH_SOURCE[0]}")"
kubectl apply -f $SCRIPT_DIR
watch kubectl describe hpa
+18
View File
@@ -0,0 +1,18 @@
#!/bin/bash
input="test-cpumanager.yaml.tmpl"
filename() {
echo "test-cpumanager-$1.yaml"
}
for runtimeclass in runc kata-qemu kata-fc; do
output=$(filename $runtimeclass)
cp $input $output
sed -i "s/__runtimeclass__/$runtimeclass/g" $output
if [ $runtimeclass == "runc" ]; then continue; fi
insertline="\ \ runtimeClassName: $runtimeclass"
sed -i "/spec:/a $insertline" $output
done
kubectl apply -f .
@@ -0,0 +1,92 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-guaranteed-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
limits:
cpu: 1
memory: 500Mi # For kata to run
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-burstable-integer-limit-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
requests:
cpu: 1
memory: 100Mi
limits:
cpu: 2
memory: 500Mi
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-burstable-float-limit-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
requests:
cpu: 500m
memory: 100Mi
limits:
cpu: 1
memory: 500Mi
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-burstable-integer-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
requests:
cpu: 1
memory: 100Mi
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-burstable-float-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
resources:
requests:
cpu: 500m
memory: 100Mi
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-besteffort-__runtimeclass__
spec:
restartPolicy: Never
containers:
- name: busy
image: busybox
command: [ "top" ]
@@ -2,19 +2,19 @@ apiVersion: apps/v1
kind: Deployment
metadata:
labels:
run: php-apache-fire
name: php-apache-fire
run: php-apache-kata-fc
name: php-apache-kata-fc
spec:
replicas: 1
selector:
matchLabels:
run: php-apache-fire
run: php-apache-kata-fc
template:
metadata:
labels:
run: php-apache-fire
run: php-apache-kata-fc
spec:
runtimeClassName: fire
runtimeClassName: kata-fc
containers:
- image: k8s.gcr.io/hpa-example
imagePullPolicy: Always
@@ -30,13 +30,13 @@ spec:
apiVersion: v1
kind: Service
metadata:
name: php-apache-fire
name: php-apache-kata-fc
spec:
ports:
- port: 80
protocol: TCP
targetPort: 80
selector:
run: php-apache-fire
run: php-apache-kata-fc
sessionAffinity: None
type: ClusterIP
@@ -2,22 +2,19 @@ apiVersion: apps/v1
kind: Deployment
metadata:
labels:
run: php-apache-kata
name: php-apache-kata
run: php-apache-kata-qemu
name: php-apache-kata-qemu
spec:
replicas: 1
selector:
matchLabels:
run: php-apache-kata
run: php-apache-kata-qemu
template:
metadata:
annotations:
io.kubernetes.cri-o.TrustedSandbox: "false"
io.kubernetes.cri.untrusted-workload: "true"
labels:
run: php-apache-kata
run: php-apache-kata-qemu
spec:
runtimeClassName: kata
runtimeClassName: kata-qemu
containers:
- image: k8s.gcr.io/hpa-example
imagePullPolicy: Always
@@ -33,13 +30,13 @@ spec:
apiVersion: v1
kind: Service
metadata:
name: php-apache-kata
name: php-apache-kata-qemu
spec:
ports:
- port: 80
protocol: TCP
targetPort: 80
selector:
run: php-apache-kata
run: php-apache-kata-qemu
sessionAffinity: None
type: ClusterIP
-6
View File
@@ -1,6 +0,0 @@
kubectl run php-apache-test --image=k8s.gcr.io/hpa-example --requests=cpu=200m --expose --port=80
kubectl autoscale deployment php-apache-test --cpu-percent=50 --min=1 --max=10
kubectl get hpa
#kubectl run -i --tty load-generator --image=busybox /bin/sh
# while true; do wget -q -O- http://php-apache-test.default.svc.cluster.local; done
@@ -1,16 +0,0 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-kata
spec:
runtimeClassName: kata
restartPolicy: Never
containers:
- name: taskset
image: busybox
command: [ "taskset", "-p", "1" ]
resources:
limits:
cpu: 1
memory: 500Mi
@@ -1,15 +0,0 @@
---
apiVersion: v1
kind: Pod
metadata:
name: test-cpumanager-runc
spec:
restartPolicy: Never
containers:
- name: taskset
image: busybox
command: [ "taskset", "-p", "1" ]
resources:
limits:
cpu: 1
memory: 100Mi