Refactor ceph to allow monitors to leverage a stateful set.

The IPs of ceph monitors cannot (or should not) change.

This refactor allows the ceph monitors to act as statefulsets.
It also persists their on disk data to nodeDirs, to allow cluster
wide restarts (in parallel), where previously this would lose
data.  This is accompanied by some docker images changes that
ensure that auth for the OSDs is restored/reinserted as auth
data is somehow not persisted to disk at this time.
This commit is contained in:
Alan Meadows 2017-01-03 12:23:25 -08:00
parent bcf23c7399
commit b366dc252a
9 changed files with 378 additions and 316 deletions

View File

@ -78,8 +78,8 @@ spec:
timeoutSeconds: 5
resources:
requests:
memory: "512Mi"
cpu: "1000m"
memory: {{ .Values.resources.osd.requests.memory | quote }}
cpu: {{ .Values.resources.osd.requests.cpu | quote }}
limits:
memory: "1024Mi"
cpu: "2000m"
memory: {{ .Values.resources.osd.limits.memory | quote }}
cpu: {{ .Values.resources.osd.limits.cpu | quote }}

View File

@ -0,0 +1,72 @@
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: mds
name: ceph-mds
spec:
replicas: 1
template:
metadata:
name: ceph-mds
labels:
app: ceph
daemon: mds
spec:
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
ports:
- containerPort: 6800
env:
- name: CEPH_DAEMON
value: MDS
- name: CEPHFS_CREATE
value: "1"
- name: KV_TYPE
value: k8s
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
livenessProbe:
tcpSocket:
port: 6800
initialDelaySeconds: 60
timeoutSeconds: 5
readinessProbe:
tcpSocket:
port: 6800
timeoutSeconds: 5
resources:
requests:
memory: {{ .Values.resources.mds.requests.memory | quote }}
cpu: {{ .Values.resources.mds.requests.cpu | quote }}
limits:
memory: {{ .Values.resources.mds.limits.memory | quote }}
cpu: {{ .Values.resources.mds.limits.cpu | quote }}

View File

@ -0,0 +1,64 @@
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: moncheck
name: ceph-mon-check
spec:
replicas: {{ .Values.replicas.mon_check }}
template:
metadata:
name: ceph-mon
labels:
app: ceph
daemon: moncheck
spec:
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
imagePullPolicy: Always
ports:
- containerPort: 6789
env:
- name: CEPH_DAEMON
value: MON_HEALTH
- name: KV_TYPE
value: k8s
- name: MON_IP_AUTO_DETECT
value: "1"
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
resources:
requests:
memory: {{ .Values.resources.mon_check.requests.memory | quote }}
cpu: {{ .Values.resources.mon_check.requests.cpu | quote }}
limits:
memory: {{ .Values.resources.mon_check.limits.memory | quote }}
cpu: {{ .Values.resources.mon_check.limits.cpu | quote }}

View File

@ -0,0 +1,77 @@
{{- if .Values.rgw.enabled }}
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: rgw
name: ceph-rgw
spec:
replicas: {{ .Values.replicas.rgw }}
template:
metadata:
name: ceph-rgw
labels:
app: ceph
daemon: rgw
spec:
hostNetwork: true
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-rgw
image: {{ .Values.images.daemon }}
ports:
- containerPort: {{ .Values.network.port.rgw_target }}
env:
- name: RGW_CIVETWEB_PORT
value: "{{ .Values.network.port.rgw_target }}"
- name: CEPH_DAEMON
value: RGW
- name: KV_TYPE
value: k8s
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
livenessProbe:
httpGet:
path: /
port: {{ .Values.network.port.rgw_target }}
initialDelaySeconds: 120
timeoutSeconds: 5
readinessProbe:
httpGet:
path: /
port: {{ .Values.network.port.rgw_target }}
timeoutSeconds: 5
resources:
requests:
memory: {{ .Values.resources.rgw.requests.memory | quote }}
cpu: {{ .Values.resources.rgwrequests.cpu | quote }}
limits:
memory: {{ .Values.resources.rgw.limits.memory | quote }}
cpu: {{ .Values.resources.rgw.limits.cpu | quote }}
{{- end }}

View File

@ -1,310 +0,0 @@
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: mds
name: ceph-mds
spec:
replicas: 1
template:
metadata:
name: ceph-mds
labels:
app: ceph
daemon: mds
spec:
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
ports:
- containerPort: 6800
env:
- name: CEPH_DAEMON
value: MDS
- name: CEPHFS_CREATE
value: "1"
- name: KV_TYPE
value: k8s
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
livenessProbe:
tcpSocket:
port: 6800
initialDelaySeconds: 60
timeoutSeconds: 5
readinessProbe:
tcpSocket:
port: 6800
timeoutSeconds: 5
resources:
requests:
memory: "10Mi"
cpu: "250m"
limits:
memory: "50Mi"
cpu: "500m"
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: moncheck
name: ceph-mon-check
spec:
replicas: 1
template:
metadata:
name: ceph-mon
labels:
app: ceph
daemon: moncheck
spec:
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
imagePullPolicy: Always
ports:
- containerPort: 6789
env:
- name: CEPH_DAEMON
value: MON_HEALTH
- name: KV_TYPE
value: k8s
- name: MON_IP_AUTO_DETECT
value: "1"
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
resources:
requests:
memory: "5Mi"
cpu: "250m"
limits:
memory: "50Mi"
cpu: "500m"
---
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: mon
name: ceph-mon
spec:
replicas: 3
template:
metadata:
name: ceph-mon
labels:
app: ceph
daemon: mon
annotations:
# alanmeadows: this soft requirement allows single
# host deployments to spawn several ceph-mon
# containers
scheduler.alpha.kubernetes.io/affinity: >
{
"podAntiAffinity": {
"preferredDuringSchedulingIgnoredDuringExecution": [{
"labelSelector": {
"matchExpressions": [{
"key": "daemon",
"operator": "In",
"values":["mon"]
}]
},
"topologyKey": "kubernetes.io/hostname",
"weight": 10
}]
}
}
spec:
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
# imagePullPolicy: Always
lifecycle:
preStop:
exec:
# remove the mon on Pod stop.
command:
- "/remove-mon.sh"
ports:
- containerPort: 6789
env:
- name: CEPH_DAEMON
value: MON
- name: KV_TYPE
value: k8s
- name: NETWORK_AUTO_DETECT
value: "1"
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
livenessProbe:
tcpSocket:
port: 6789
initialDelaySeconds: 60
timeoutSeconds: 5
readinessProbe:
tcpSocket:
port: 6789
timeoutSeconds: 5
resources:
requests:
memory: "50Mi"
cpu: "1000m"
limits:
memory: "100Mi"
cpu: "2000m"
---
# rgw not required: using if statement for deployment
{{- if .Values.rgw.enabled }}
kind: Deployment
apiVersion: extensions/v1beta1
metadata:
labels:
app: ceph
daemon: rgw
name: ceph-rgw
spec:
replicas: 3
template:
metadata:
name: ceph-rgw
labels:
app: ceph
daemon: rgw
spec:
hostNetwork: true
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
containers:
- name: ceph-rgw
image: {{ .Values.images.daemon }}
ports:
- containerPort: {{ .Values.network.port.rgw_target }}
env:
- name: RGW_CIVETWEB_PORT
value: "{{ .Values.network.port.rgw_target }}"
- name: CEPH_DAEMON
value: RGW
- name: KV_TYPE
value: k8s
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
livenessProbe:
httpGet:
path: /
port: {{ .Values.network.port.rgw_target }}
initialDelaySeconds: 120
timeoutSeconds: 5
readinessProbe:
httpGet:
path: /
port: {{ .Values.network.port.rgw_target }}
timeoutSeconds: 5
resources:
requests:
memory: "500Mi"
cpu: ".5"
limits:
memory: "500Mi"
cpu: ".5"
{{ end }}
# end: rgw removed optionally

View File

@ -15,6 +15,8 @@ spec:
app: ceph
daemon: mon
clusterIP: None
{{- if .Values.rgw.enabled }}
---
apiVersion: v1
kind: Service
@ -32,3 +34,4 @@ spec:
app: ceph
daemon: rgw
type: LoadBalancer
{{- end }}

View File

@ -0,0 +1,105 @@
---
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
labels:
app: ceph
daemon: mon
name: ceph-mon
spec:
serviceName: {{ .Values.service.mon.name | quote }}
replicas: {{ .Values.replicas.mon }}
template:
metadata:
name: ceph-mon
labels:
app: ceph
daemon: mon
annotations:
# alanmeadows: this soft requirement allows single
# host deployments to spawn several ceph-mon
# containers
scheduler.alpha.kubernetes.io/affinity: >
{
"podAntiAffinity": {
"preferredDuringSchedulingIgnoredDuringExecution": [{
"labelSelector": {
"matchExpressions": [{
"key": "daemon",
"operator": "In",
"values":["mon"]
}]
},
"topologyKey": "kubernetes.io/hostname",
"weight": 10
}]
}
}
spec:
nodeSelector:
{{ .Values.labels.node_selector_key }}: {{ .Values.labels.node_selector_value }}
serviceAccount: default
volumes:
- name: ceph-conf
secret:
secretName: ceph-conf-combined
- name: ceph-bootstrap-osd-keyring
secret:
secretName: ceph-bootstrap-osd-keyring
- name: ceph-bootstrap-mds-keyring
secret:
secretName: ceph-bootstrap-mds-keyring
- name: ceph-bootstrap-rgw-keyring
secret:
secretName: ceph-bootstrap-rgw-keyring
- name: ceph-monfs
hostPath:
path: {{ .Values.storage.mon_directory }}
containers:
- name: ceph-mon
image: {{ .Values.images.daemon }}
imagePullPolicy: Always
lifecycle:
preStop:
exec:
# remove the mon on Pod stop.
command:
- "/remove-mon.sh"
ports:
- containerPort: 6789
env:
- name: CEPH_DAEMON
value: MON
- name: KV_TYPE
value: k8s
- name: NETWORK_AUTO_DETECT
value: "1"
- name: CLUSTER
value: ceph
volumeMounts:
- name: ceph-conf
mountPath: /etc/ceph
- name: ceph-bootstrap-osd-keyring
mountPath: /var/lib/ceph/bootstrap-osd
- name: ceph-bootstrap-mds-keyring
mountPath: /var/lib/ceph/bootstrap-mds
- name: ceph-bootstrap-rgw-keyring
mountPath: /var/lib/ceph/bootstrap-rgw
- name: ceph-monfs
mountPath: /var/lib/ceph/mon
livenessProbe:
tcpSocket:
port: 6789
initialDelaySeconds: 60
timeoutSeconds: 5
readinessProbe:
tcpSocket:
port: 6789
timeoutSeconds: 5
resources:
requests:
memory: {{ .Values.resources.mon.requests.memory | quote }}
cpu: {{ .Values.resources.mon.requests.cpu | quote }}
limits:
memory: {{ .Values.resources.mon.limits.memory | quote }}
cpu: {{ .Values.resources.mon.limits.cpu | quote }}

View File

@ -10,7 +10,8 @@ parameters:
monitors: {{ .Values.storageclass.monitors | default "ceph-mon.ceph:6789" }}
adminId: {{ .Values.storageclass.admin_id }}
adminSecretName: {{ .Values.storageclass.admin_secret_name }}
## forcing namespace due to issue with -- default "{{ .Release.Namespace }}" }} --
# forcing namespace due to issue with default pipeline of "{{ .Release.Namespace }}" }}
# during helm lint
adminSecretNamespace: {{ .Values.storageclass.admin_secret_namespace | default "ceph" }}
pool: {{ .Values.storageclass.pool }}
userId: {{ .Values.storageclass.user_id }}

View File

@ -7,6 +7,15 @@
# tunables available - parameterizing more of the elements
# in the manifests is a work in progress
replicas:
mon: 3
rgw: 3
mon_check: 1
service:
mon:
name: ceph-mon
images:
daemon: quay.io/attcomdev/ceph-daemon:latest
@ -23,11 +32,52 @@ network:
storage:
osd_directory: /var/lib/openstack-helm/ceph/osd
var_directory: /var/lib/openstack-helm/ceph/ceph
mon_directory: /var/lib/openstack-helm/ceph/mon
# rgw is optionall disabled
rgw:
enabled: false
rgw:
enabled: false
resources:
osd:
requests:
memory: "512Mi"
cpu: "1000m"
limits:
memory: "1024Mi"
cpu: "2000m"
mds:
requests:
memory: "10Mi"
cpu: "250m"
limits:
memory: "50Mi"
cpu: "500m"
mon:
requests:
memory: "50Mi"
cpu: "1000m"
limits:
memory: "100Mi"
cpu: "2000m"
mon_check:
requests:
memory: "5Mi"
cpu: "250m"
limits:
memory: "50Mi"
cpu: "500m"
rgw:
requests:
memory: "5Mi"
cpu: "250m"
limits:
memory: "50Mi"
cpu: "500m"
# Setting this to false will assume you will
# setup and orchestrate your own secrets and
# configmaps outside of this helm chart
@ -63,7 +113,7 @@ secrets:
storageclass:
provision_storage_class: true
name: general
monitors: null
monitors: null
pool: rbd
admin_id: admin
admin_secret_name: pvc-ceph-conf-combined-storageclass