Merging master into branch

.goreleaser.yml

@@ -195,7 +195,7 @@ dockers:
       - --label=org.opencontainers.image.version={{ .Version }}
       - --label=org.opencontainers.image.created={{ time "2006-01-02T15:04:05Z07:00" }}
       - --label=org.opencontainers.image.revision={{ .FullCommit }}
-      - --label=org.opencontainers.image.base.name=alpine:3.18.3
+      - --label=org.opencontainers.image.base.name=alpine:3.20.1
       - --label=org.opencontainers.image.licenses=Apache-2.0
       - --label=org.opencontainers.image.vendor=G-Research
     ids:

build/armada-load-tester/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/armada/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/armadactl/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/binoculars/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/eventingester/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/executor/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/fakeexecutor/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/jobservice/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/lookoutingesterv2/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/lookoutv2/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/scheduler/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/scheduleringester/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build/testsuite/Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.18.3
+FROM alpine:3.20.1
 
 RUN addgroup -S -g 2000 armada && adduser -S -u 1000 armada -G armada
 

build_goreleaser/armadactl/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 FROM ${BASE_IMAGE}
 
 LABEL org.opencontainers.image.title=armadactl

build_goreleaser/binoculars/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=binoculars
 LABEL org.opencontainers.image.description="binoculars"

build_goreleaser/bundles/armada/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=armada
 LABEL org.opencontainers.image.description="Armada Bundle"

build_goreleaser/bundles/full/Dockerfile

@@ -1,6 +1,6 @@
 ARG NODE_BUILD_IMAGE=node:16.14-buster
 ARG OPENAPI_BUILD_IMAGE=openapitools/openapi-generator-cli:v5.4.0
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${OPENAPI_BUILD_IMAGE} AS OPENAPI
 LABEL org.opencontainers.image.title=armada-full-bundle

build_goreleaser/bundles/lookout/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=armada-lookout-bundle

build_goreleaser/eventingester/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.5
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=eventingester

build_goreleaser/executor/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=executor

build_goreleaser/fakeexecutor/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=fakeexecutor
 LABEL org.opencontainers.image.description="Fake Executor"

build_goreleaser/jobservice/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=jobservice

build_goreleaser/loadtester/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=loadtester

build_goreleaser/lookoutingesterv2/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=lookoutingesterv2
@@ -13,4 +13,4 @@ COPY config/lookoutingesterv2/config.yaml /app/config/lookoutingesterv2/config.y
 
 WORKDIR /app
 
-ENTRYPOINT ["./lookoutingesterv2"]
+ENTRYPOINT ["./lookoutingesterv2"]

build_goreleaser/lookoutv2/Dockerfile

@@ -1,6 +1,6 @@
 ARG NODE_BUILD_IMAGE=node:16.14-buster
 ARG OPENAPI_BUILD_IMAGE=openapitools/openapi-generator-cli:v5.4.0
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${OPENAPI_BUILD_IMAGE} AS OPENAPI
 

build_goreleaser/scheduler/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=scheduler

build_goreleaser/scheduleringester/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=scheduleringester

build_goreleaser/server/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=armada-server

build_goreleaser/testsuite/Dockerfile

@@ -1,4 +1,4 @@
-ARG BASE_IMAGE=alpine:3.18.3
+ARG BASE_IMAGE=alpine:3.20.1
 FROM ${BASE_IMAGE}
 LABEL org.opencontainers.image.title=testsuite
 LABEL org.opencontainers.image.description="Test Suite"

cmd/armada-load-tester/cmd/loadtest.go

@@ -52,7 +52,7 @@ var loadtestCmd = &cobra.Command{
 			  containers:
 				- name: sleep
 				  imagePullPolicy: IfNotPresent
-				  image: alpine:3.18.3
+				  image: alpine:3.20.1
 				  command:
 					- sh
 				  args:

config/scheduler/config.yaml

@@ -87,6 +87,7 @@ scheduling:
   disableScheduling: false
   enableAssertions: false
   protectedFractionOfFairShare: 1.0
+  useAdjustedFairShareProtection: true
   nodeIdLabel: "kubernetes.io/hostname"
   priorityClasses:
     armada-default:

developer/config/job.yaml

@@ -11,7 +11,7 @@ jobs:
       containers:
         - name: sleep
           imagePullPolicy: IfNotPresent
-          image: alpine:3.10
+          image: alpine:latest
           args:
             - "exit"
             - "1"
@@ -26,4 +26,4 @@ jobs:
 timeout: "100s"
 expectedEvents:
   - submitted:
-  - failed:
+  - failed:

e2e/armadactl_test/armadactl_test.go

@@ -175,7 +175,7 @@ jobs:
       containers:
         - name: ls
           imagePullPolicy: IfNotPresent
-          image: alpine:3.18.3
+          image: alpine:3.20.1
           command:
             - sh
             - -c

internal/armada/configuration/types.go

@@ -125,6 +125,9 @@ type SubmissionConfig struct {
 	// will have activeDeadlineSeconds set to 1.
 	// Trumps DefaultActiveDeadline.
 	DefaultActiveDeadlineByResourceRequest map[string]time.Duration
+	// Maximum ratio of limits:requests per resource. Jobs who have a higher limits:resource ratio than this will be rejected.
+	// Any resource type missing from this map will default to 1.0.
+	MaxOversubscriptionByResourceRequest map[string]float64
 }
 
 // TODO: we can probably just typedef this to map[string]string

internal/armada/queryapi/query_api.go

@@ -24,6 +24,7 @@ var JobStateMap = map[int16]api.JobState{
 	lookout.JobFailedOrdinal:    api.JobState_FAILED,
 	lookout.JobCancelledOrdinal: api.JobState_CANCELLED,
 	lookout.JobPreemptedOrdinal: api.JobState_PREEMPTED,
+	lookout.JobRejectedOrdinal:  api.JobState_REJECTED,
 }
 
 // JobRunStateMap is a mapping between database state and api Job Run states

internal/armada/submit/validation/submit_request.go

@@ -3,8 +3,6 @@ package validation
 import (
 	"fmt"
 
-	v1 "k8s.io/api/core/v1"
-
 	"github.com/pkg/errors"
 	"k8s.io/component-helpers/scheduling/corev1/nodeaffinity"
 
@@ -210,28 +208,36 @@ func validatePriorityClasses(j *api.JobSubmitRequestItem, config configuration.S
 // Ensures that the JobSubmitRequestItem's limits and requests are equal.
 // Also  checks that  any resources defined are above minimum values set in  config
 func validateResources(j *api.JobSubmitRequestItem, config configuration.SubmissionConfig) error {
-	// Function which tells us if two k8s resource lists contain exactly the same elements
-	resourceListEquals := func(a v1.ResourceList, b v1.ResourceList) bool {
-		if len(a) != len(b) {
-			return false
-		}
-		for k, v := range a {
-			if v != b[k] {
-				return false
-			}
-		}
-		return true
-	}
-
 	spec := j.GetMainPodSpec()
+	maxOversubscriptionByResource := config.MaxOversubscriptionByResourceRequest
+	if maxOversubscriptionByResource == nil {
+		maxOversubscriptionByResource = map[string]float64{}
+	}
 	for _, container := range spec.Containers {
 
 		if len(container.Resources.Requests) == 0 && len(container.Resources.Requests) == 0 {
 			return fmt.Errorf("container %v has no resources specified", container.Name)
 		}
 
-		if !resourceListEquals(container.Resources.Requests, container.Resources.Limits) {
-			return fmt.Errorf("container %v does not have resource request and limit equal (this is currently not supported)", container.Name)
+		if len(container.Resources.Requests) != len(container.Resources.Limits) {
+			return fmt.Errorf("container %v defines different resources for requests and limits", container.Name)
+		}
+
+		for resource, request := range container.Resources.Requests {
+			limit, ok := container.Resources.Limits[resource]
+			if !ok {
+				return fmt.Errorf("container %v defines %s for requests but not limits", container.Name, resource)
+			}
+			if limit.MilliValue() < request.MilliValue() {
+				return fmt.Errorf("container %v defines %s with limits smaller than requests", container.Name, resource)
+			}
+			maxOversubscription, ok := maxOversubscriptionByResource[resource.String()]
+			if !ok {
+				maxOversubscription = 1.0
+			}
+			if float64(limit.MilliValue()) > maxOversubscription*float64(request.MilliValue()) {
+				return fmt.Errorf("container %v defines %s with limits great than %.2f*requests", container.Name, resource, maxOversubscription)
+			}
 		}
 
 		for rc, containerRsc := range container.Resources.Requests {