Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Appearance settings

Commit 8edb9dc

Browse filesBrowse files
ppaluckippalucki
committed
VPA: v1 - first version of vertical pod autoscaler
1 parent acb9ea1 commit 8edb9dc
Copy full SHA for 8edb9dc

File tree

6 files changed

+103
-8
lines changed
Filter options

6 files changed

+103
-8
lines changed

‎.dockerignore

Copy file name to clipboard
+1Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
/deployment

‎.gitignore

Copy file name to clipboardExpand all lines: .gitignore
+1Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -36,3 +36,4 @@ src/simdjson
3636
/deployment/pcm/smarter-device-manager/
3737
/deployment/pcm/nri/
3838
/deployment/pcm/kind-with-registry.sh
39+
/deployment/pcm/autoscaler

‎deployment/pcm/.helmignore

Copy file name to clipboardExpand all lines: deployment/pcm/.helmignore
+1Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -23,3 +23,4 @@
2323
.vscode/
2424
smarter-device-manager/
2525
nri/
26+
autoscaler/

‎deployment/pcm/README.md

Copy file name to clipboardExpand all lines: deployment/pcm/README.md
+28-8Lines changed: 28 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -4,13 +4,14 @@ Helm chart instructions
44

55
### Features:
66

7-
- Configurable as non-privileged container (value: `privileged=false`, default) and privileged container,
7+
- Configurable as non-privileged container (value: `privileged`, default: false) and privileged container,
88
- Support for bare-metal and VM host configurations (files: [values-metal.yaml](values-metal.yaml), [values-vm.yaml](values-vm.yaml)),
99
- Ability to deploy multiple releases alongside configured differently to handle different kinds of machines (bare-metal, VM) at the [same time](#heterogeneous-mixed-vmmetal-instances-cluster),
1010
- Linux Watchdog handling (controlled with `PCM_KEEP_NMI_WATCHDOG`, `PCM_NO_AWS_WORKAROUND`, `nmiWatchdogMount` values).
1111
- Deploy to own namespace with "helm install ... **-n pcm --create-namespace**".
12-
- Silent mode (value: `silent=false`, default).
13-
- Backward compatbile with older Linux kernels (<5.8) - (value: cap_perfmon).
12+
- Silent mode (value: `silent`, default: false).
13+
- Backward compatible with older Linux kernels (<5.8) - (value: cap_perfmon, default: false).
14+
- VerticalPodAutoscaler (value: `verticalPodAutoscaler.enabled`, default: false)
1415

1516
Here are available methods in this chart of metrics collection w.r.t interfaces and required access:
1617

@@ -47,13 +48,12 @@ helm install pcm .
4748
helm install pcm . -f values-direct-privileged.yaml
4849
```
4950

50-
#### Node-feature-discovery + Prometheus podMonitor
51+
#### All opt-in features: Node-feature-discovery + Prometheus podMonitor + vertical
5152

5253
```
53-
helm install ... --set nfd=true --set podMonitor=true
54+
helm install ... --set nfd=true --set podMonitor=true --set verticalPodAutoscaler.enabled=true
5455
```
5556

56-
5757
### Requirements
5858

5959
- Full set of metrics (uncore/UPI, RDT, energy) requires bare-metal or .metal cloud instance.
@@ -180,7 +180,27 @@ kubectl get sts prometheus-prometheus-kube-prometheus-prometheus
180180
Note: `podMonitorSelectorNilUsesHelmValues` is disabled (set to false) so Prometheus operator will be able to handle PCM podMonitor deployed without extra `podMonitorLabels` or otherwise pcm need to be deployed like this:
181181
`helm install pcm . --set podMonitor=true --set podMonitorLabels.release=prometheus` (assuming Prometheus operator was deployed as "prometheus")
182182

183-
#### 5) Deploy PCM helm chart
183+
184+
#### 5) (Optionally) Deploy metric-server and vertical-pod-autoscaler
185+
186+
Note this is irrelevant to pcm-sensor-server functionality, but useful to observer pcm pod CPU/memory usage:
187+
188+
a) metric-server
189+
190+
```
191+
helm repo add metrics-server https://kubernetes-sigs.github.io/metrics-server/
192+
helm repo update
193+
helm upgrade --install --set args={--kubelet-insecure-tls} metrics-server metrics-server/metrics-server --namespace kube-system
194+
```
195+
196+
b) vertical pod autoscaler
197+
198+
```
199+
git clone https://github.com/kubernetes/autoscaler
200+
./autoscaler/vertical-pod-autoscaler/hack/vpa-up.sh
201+
```
202+
203+
#### 6) Deploy PCM helm chart
184204

185205
```
186206
# a) Deploy to current namespace with defaults
@@ -194,7 +214,7 @@ helm install pcm . --set nfd=true
194214
helm install pcm . --namespace pcm
195215
```
196216

197-
#### 6) Check metrics are exported
217+
#### 7) Check metrics are exported
198218

199219
Run proxy in background:
200220
```

‎deployment/pcm/templates/verticalpodautoscaler.yaml

Copy file name to clipboard
+40Lines changed: 40 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,40 @@
1+
{{- if and (.Capabilities.APIVersions.Has "autoscaling.k8s.io/v1") (.Values.verticalPodAutoscaler.enabled) }}
2+
apiVersion: autoscaling.k8s.io/v1
3+
kind: VerticalPodAutoscaler
4+
metadata:
5+
name: {{ include "pcm.fullname" . }}
6+
namespace: {{ .Release.Namespace }}
7+
labels:
8+
{{- include "pcm.labels" . | nindent 4 }}
9+
spec:
10+
{{- with .Values.verticalPodAutoscaler.recommenders }}
11+
recommenders:
12+
{{- toYaml . | nindent 4 }}
13+
{{- end }}
14+
resourcePolicy:
15+
containerPolicies:
16+
- containerName: pcm
17+
{{- with .Values.verticalPodAutoscaler.controlledResources }}
18+
controlledResources:
19+
{{- toYaml . | nindent 8 }}
20+
{{- end }}
21+
{{- with .Values.verticalPodAutoscaler.controlledValues }}
22+
controlledValues: {{ . }}
23+
{{- end }}
24+
{{- with .Values.verticalPodAutoscaler.maxAllowed }}
25+
maxAllowed:
26+
{{- toYaml . | nindent 8 }}
27+
{{- end }}
28+
{{- with .Values.verticalPodAutoscaler.minAllowed }}
29+
minAllowed:
30+
{{- toYaml . | nindent 8 }}
31+
{{- end }}
32+
targetRef:
33+
apiVersion: apps/v1
34+
kind: DaemonSet
35+
name: {{ include "pcm.fullname" . }}
36+
{{- with .Values.verticalPodAutoscaler.updatePolicy }}
37+
updatePolicy:
38+
{{- toYaml . | nindent 4 }}
39+
{{- end }}
40+
{{- end }}

‎deployment/pcm/values.yaml

Copy file name to clipboardExpand all lines: deployment/pcm/values.yaml
+32Lines changed: 32 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -131,3 +131,35 @@ nfdBaremetalAffinity: false
131131
# feature.node.kubernetes.io/cpu-rdt.RDTMON=true
132132
nfdRDTAffinity: false
133133

134+
135+
### -------------- verticalPodAutoscaler ------------------
136+
# Enable vertical pod autoscaler support for pcm-sensor-server
137+
verticalPodAutoscaler:
138+
enabled: false
139+
140+
# Recommender responsible for generating recommendation for the object.
141+
# List should be empty (then the default recommender will generate the recommendation)
142+
# or contain exactly one recommender.
143+
# recommenders:
144+
# - name: custom-recommender-performance
145+
146+
# List of resources that the vertical pod autoscaler can control. Defaults to cpu and memory
147+
controlledResources: []
148+
# Specifies which resource values should be controlled: RequestsOnly or RequestsAndLimits.
149+
# controlledValues: RequestsAndLimits
150+
151+
# Define the max allowed resources for the pod
152+
maxAllowed: {}
153+
# cpu: 200m
154+
# memory: 100Mi
155+
# Define the min allowed resources for the pod
156+
minAllowed: {}
157+
# cpu: 200m
158+
# memory: 100Mi
159+
160+
# updatePolicy:
161+
# Specifies minimal number of replicas which need to be alive for VPA Updater to attempt pod eviction
162+
# minReplicas: 1
163+
# Specifies whether recommended updates are applied when a Pod is started and whether recommended updates
164+
# are applied during the life of a Pod. Possible values are "Off", "Initial", "Recreate", and "Auto".
165+
# updateMode: Auto

0 commit comments

Comments
0 (0)
Morty Proxy This is a proxified and sanitized view of the page, visit original site.