mirror of
https://github.com/ceph/ceph-csi.git
synced 2024-11-14 02:10:21 +00:00
139 lines
6.7 KiB
Bash
139 lines
6.7 KiB
Bash
#!/bin/bash
|
|
|
|
# Copyright 2017 The Kubernetes Authors.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
|
|
# A configuration for Kubemark cluster. It doesn't need to be kept in
|
|
# sync with gce/config-default.sh (except the filename, because I'm reusing
|
|
# gce/util.sh script which assumes config filename), but if some things that
|
|
# are enabled by default should not run in hollow clusters, they should be disabled here.
|
|
|
|
source "${KUBE_ROOT}/cluster/gce/config-common.sh"
|
|
|
|
GCLOUD=gcloud
|
|
ZONE=${KUBE_GCE_ZONE:-us-central1-b}
|
|
REGION=${ZONE%-*}
|
|
NUM_NODES=${KUBEMARK_NUM_NODES:-10}
|
|
MASTER_SIZE=${KUBEMARK_MASTER_SIZE:-n1-standard-$(get-master-size)}
|
|
MASTER_DISK_TYPE=pd-ssd
|
|
MASTER_DISK_SIZE=${MASTER_DISK_SIZE:-$(get-master-disk-size)}
|
|
MASTER_ROOT_DISK_SIZE=${KUBEMARK_MASTER_ROOT_DISK_SIZE:-$(get-master-root-disk-size)}
|
|
REGISTER_MASTER_KUBELET=${REGISTER_MASTER:-false}
|
|
PREEMPTIBLE_NODE=${PREEMPTIBLE_NODE:-false}
|
|
NODE_ACCELERATORS=${NODE_ACCELERATORS:-""}
|
|
CREATE_CUSTOM_NETWORK=${CREATE_CUSTOM_NETWORK:-false}
|
|
EVENT_PD=${EVENT_PD:-false}
|
|
|
|
MASTER_OS_DISTRIBUTION=${KUBE_MASTER_OS_DISTRIBUTION:-gci}
|
|
NODE_OS_DISTRIBUTION=${KUBE_NODE_OS_DISTRIBUTION:-gci}
|
|
MASTER_IMAGE=${KUBE_GCE_MASTER_IMAGE:-cos-stable-63-10032-71-0}
|
|
MASTER_IMAGE_PROJECT=${KUBE_GCE_MASTER_PROJECT:-cos-cloud}
|
|
|
|
# GPUs supported in GCE do not have compatible drivers in Debian 7.
|
|
if [[ "${NODE_OS_DISTRIBUTION}" == "debian" ]]; then
|
|
NODE_ACCELERATORS=""
|
|
fi
|
|
|
|
NETWORK=${KUBE_GCE_NETWORK:-e2e}
|
|
if [[ "${CREATE_CUSTOM_NETWORK}" == true ]]; then
|
|
SUBNETWORK="${SUBNETWORK:-${NETWORK}-custom-subnet}"
|
|
fi
|
|
INSTANCE_PREFIX="${INSTANCE_PREFIX:-"default"}"
|
|
MASTER_NAME="${INSTANCE_PREFIX}-kubemark-master"
|
|
AGGREGATOR_MASTER_NAME="${INSTANCE_PREFIX}-kubemark-aggregator"
|
|
MASTER_TAG="kubemark-master"
|
|
ETCD_QUORUM_READ="${ENABLE_ETCD_QUORUM_READ:-false}"
|
|
EVENT_STORE_NAME="${INSTANCE_PREFIX}-event-store"
|
|
MASTER_IP_RANGE="${MASTER_IP_RANGE:-10.246.0.0/24}"
|
|
CLUSTER_IP_RANGE="${CLUSTER_IP_RANGE:-$(get-cluster-ip-range)}"
|
|
RUNTIME_CONFIG="${KUBE_RUNTIME_CONFIG:-}"
|
|
TERMINATED_POD_GC_THRESHOLD=${TERMINATED_POD_GC_THRESHOLD:-100}
|
|
KUBE_APISERVER_REQUEST_TIMEOUT=300
|
|
ETCD_COMPACTION_INTERVAL_SEC="${KUBEMARK_ETCD_COMPACTION_INTERVAL_SEC:-}"
|
|
|
|
# Set etcd image (e.g. k8s.gcr.io/etcd) and version (e.g. 3.1.10) if you need
|
|
# non-default version.
|
|
ETCD_IMAGE="${TEST_ETCD_IMAGE:-}"
|
|
ETCD_VERSION="${TEST_ETCD_VERSION:-}"
|
|
|
|
# Storage backend. 'etcd2' and 'etcd3' are supported.
|
|
STORAGE_BACKEND=${STORAGE_BACKEND:-}
|
|
# Storage media type: application/json and application/vnd.kubernetes.protobuf are supported.
|
|
STORAGE_MEDIA_TYPE=${STORAGE_MEDIA_TYPE:-}
|
|
|
|
# Default Log level for all components in test clusters and variables to override it in specific components.
|
|
TEST_CLUSTER_LOG_LEVEL="${TEST_CLUSTER_LOG_LEVEL:---v=2}"
|
|
KUBELET_TEST_LOG_LEVEL="${KUBELET_TEST_LOG_LEVEL:-$TEST_CLUSTER_LOG_LEVEL}"
|
|
API_SERVER_TEST_LOG_LEVEL="${API_SERVER_TEST_LOG_LEVEL:-$TEST_CLUSTER_LOG_LEVEL}"
|
|
CONTROLLER_MANAGER_TEST_LOG_LEVEL="${CONTROLLER_MANAGER_TEST_LOG_LEVEL:-$TEST_CLUSTER_LOG_LEVEL}"
|
|
SCHEDULER_TEST_LOG_LEVEL="${SCHEDULER_TEST_LOG_LEVEL:-$TEST_CLUSTER_LOG_LEVEL}"
|
|
KUBEPROXY_TEST_LOG_LEVEL="${KUBEPROXY_TEST_LOG_LEVEL:-$TEST_CLUSTER_LOG_LEVEL}"
|
|
|
|
TEST_CLUSTER_DELETE_COLLECTION_WORKERS="${TEST_CLUSTER_DELETE_COLLECTION_WORKERS:---delete-collection-workers=16}"
|
|
TEST_CLUSTER_MAX_REQUESTS_INFLIGHT="${TEST_CLUSTER_MAX_REQUESTS_INFLIGHT:-}"
|
|
TEST_CLUSTER_RESYNC_PERIOD="${TEST_CLUSTER_RESYNC_PERIOD:-}"
|
|
|
|
# ContentType used by all components to communicate with apiserver.
|
|
TEST_CLUSTER_API_CONTENT_TYPE="${TEST_CLUSTER_API_CONTENT_TYPE:-}"
|
|
|
|
KUBEMARK_MASTER_COMPONENTS_QPS_LIMITS="${KUBEMARK_MASTER_COMPONENTS_QPS_LIMITS:-}"
|
|
|
|
CUSTOM_ADMISSION_PLUGINS="${CUSTOM_ADMISSION_PLUGINS:-Initializers,NamespaceLifecycle,LimitRanger,ServiceAccount,PersistentVolumeLabel,PodPreset,DefaultStorageClass,DefaultTolerationSeconds,NodeRestriction,Priority,StorageObjectInUseProtection,MutatingAdmissionWebhook,ValidatingAdmissionWebhook,ResourceQuota}"
|
|
|
|
# Master components' test arguments.
|
|
APISERVER_TEST_ARGS="${KUBEMARK_APISERVER_TEST_ARGS:-} --runtime-config=extensions/v1beta1 ${API_SERVER_TEST_LOG_LEVEL} ${TEST_CLUSTER_MAX_REQUESTS_INFLIGHT} ${TEST_CLUSTER_DELETE_COLLECTION_WORKERS}"
|
|
CONTROLLER_MANAGER_TEST_ARGS="${KUBEMARK_CONTROLLER_MANAGER_TEST_ARGS:-} ${CONTROLLER_MANAGER_TEST_LOG_LEVEL} ${TEST_CLUSTER_RESYNC_PERIOD} ${TEST_CLUSTER_API_CONTENT_TYPE} ${KUBEMARK_MASTER_COMPONENTS_QPS_LIMITS}"
|
|
SCHEDULER_TEST_ARGS="${KUBEMARK_SCHEDULER_TEST_ARGS:-} ${SCHEDULER_TEST_LOG_LEVEL} ${TEST_CLUSTER_API_CONTENT_TYPE} ${KUBEMARK_MASTER_COMPONENTS_QPS_LIMITS}"
|
|
|
|
# Hollow-node components' test arguments.
|
|
KUBELET_TEST_ARGS="--max-pods=100 $TEST_CLUSTER_LOG_LEVEL ${TEST_CLUSTER_API_CONTENT_TYPE}"
|
|
KUBEPROXY_TEST_ARGS="${KUBEPROXY_TEST_LOG_LEVEL} ${TEST_CLUSTER_API_CONTENT_TYPE}"
|
|
USE_REAL_PROXIER=${USE_REAL_PROXIER:-true} # for hollow-proxy
|
|
|
|
SERVICE_CLUSTER_IP_RANGE="10.0.0.0/16" # formerly PORTAL_NET
|
|
ALLOCATE_NODE_CIDRS=true
|
|
|
|
# Optional: Enable cluster autoscaler.
|
|
ENABLE_KUBEMARK_CLUSTER_AUTOSCALER="${ENABLE_KUBEMARK_CLUSTER_AUTOSCALER:-false}"
|
|
# When using Cluster Autoscaler, always start with one hollow-node replica.
|
|
# NUM_NODES should not be specified by the user. Instead we use
|
|
# NUM_NODES=KUBEMARK_AUTOSCALER_MAX_NODES. This gives other cluster components
|
|
# (e.g. kubemark master, Heapster) enough resources to handle maximum cluster size.
|
|
if [[ "${ENABLE_KUBEMARK_CLUSTER_AUTOSCALER}" == "true" ]]; then
|
|
NUM_REPLICAS=1
|
|
if [[ ! -z "$NUM_NODES" ]]; then
|
|
echo "WARNING: Using Cluster Autoscaler, ignoring NUM_NODES parameter. Set KUBEMARK_AUTOSCALER_MAX_NODES to specify maximum size of the cluster."
|
|
fi
|
|
fi
|
|
|
|
# Optional: set feature gates
|
|
FEATURE_GATES="${KUBE_FEATURE_GATES:-ExperimentalCriticalPodAnnotation=true}"
|
|
|
|
if [[ ! -z "${NODE_ACCELERATORS}" ]]; then
|
|
FEATURE_GATES="${FEATURE_GATES},Accelerators=true"
|
|
fi
|
|
|
|
# Enable a simple "AdvancedAuditing" setup for testing.
|
|
ENABLE_APISERVER_ADVANCED_AUDIT="${ENABLE_APISERVER_ADVANCED_AUDIT:-false}"
|
|
|
|
# Optional: enable pod priority
|
|
ENABLE_POD_PRIORITY="${ENABLE_POD_PRIORITY:-}"
|
|
if [[ "${ENABLE_POD_PRIORITY}" == "true" ]]; then
|
|
FEATURE_GATES="${FEATURE_GATES},PodPriority=true"
|
|
fi
|
|
|
|
# The number of services that are allowed to sync concurrently. Will be passed
|
|
# into kube-controller-manager via `--concurrent-service-syncs`
|
|
CONCURRENT_SERVICE_SYNCS="${CONCURRENT_SERVICE_SYNCS:-}"
|