Sébastien Han
a3068dee0b
Now, the CephBlockPool CRD is managed with the controller-runtime. So the watcher is outside of the main controller reconciliation loop of CephCluster which brings numerous benefit such as: * having its own reconciliation loop * won't block anything from the main CephCluster controller loop * fast than waiting for CephCluster loop to completion Partially close: https://github.com/rook/rook/issues/1981 Signed-off-by: Sébastien Han <seb@redhat.com>
253 lines
8.1 KiB
Go
253 lines
8.1 KiB
Go
/*
|
|
Copyright 2018 The Rook Authors. All rights reserved.
|
|
|
|
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.
|
|
*/
|
|
|
|
package nfs
|
|
|
|
import (
|
|
"github.com/pkg/errors"
|
|
cephv1 "github.com/rook/rook/pkg/apis/ceph.rook.io/v1"
|
|
cephconfig "github.com/rook/rook/pkg/daemon/ceph/config"
|
|
"github.com/rook/rook/pkg/operator/ceph/cluster/mon"
|
|
"github.com/rook/rook/pkg/operator/ceph/config/keyring"
|
|
"github.com/rook/rook/pkg/operator/ceph/controller"
|
|
"github.com/rook/rook/pkg/operator/k8sutil"
|
|
apps "k8s.io/api/apps/v1"
|
|
v1 "k8s.io/api/core/v1"
|
|
kerrors "k8s.io/apimachinery/pkg/api/errors"
|
|
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
|
|
"k8s.io/apimachinery/pkg/util/intstr"
|
|
)
|
|
|
|
const (
|
|
appName = "rook-ceph-nfs"
|
|
ganeshaConfigVolume = "ganesha-config"
|
|
nfsPort = 2049
|
|
ganeshaPid = "/var/run/ganesha/ganesha.pid"
|
|
)
|
|
|
|
func (c *CephNFSController) createCephNFSService(nfs cephv1.CephNFS, cfg daemonConfig) error {
|
|
labels := getLabels(nfs, cfg.ID)
|
|
svc := &v1.Service{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: instanceName(nfs, cfg.ID),
|
|
Namespace: nfs.Namespace,
|
|
Labels: labels,
|
|
OwnerReferences: ownerRefs(nfs),
|
|
},
|
|
Spec: v1.ServiceSpec{
|
|
Selector: labels,
|
|
Ports: []v1.ServicePort{
|
|
{
|
|
Name: "nfs",
|
|
Port: nfsPort,
|
|
TargetPort: intstr.FromInt(int(nfsPort)),
|
|
Protocol: v1.ProtocolTCP,
|
|
},
|
|
},
|
|
},
|
|
}
|
|
if c.clusterSpec.Network.IsHost() {
|
|
svc.Spec.ClusterIP = v1.ClusterIPNone
|
|
}
|
|
|
|
svc, err := c.context.Clientset.CoreV1().Services(nfs.Namespace).Create(svc)
|
|
if err != nil {
|
|
if !kerrors.IsAlreadyExists(err) {
|
|
return errors.Wrapf(err, "failed to create ganesha service")
|
|
}
|
|
logger.Infof("ceph nfs service already created")
|
|
return nil
|
|
}
|
|
|
|
logger.Infof("ceph nfs service running at %s:%d", svc.Spec.ClusterIP, nfsPort)
|
|
return nil
|
|
}
|
|
|
|
func (c *CephNFSController) makeDeployment(nfs cephv1.CephNFS, cfg daemonConfig) *apps.Deployment {
|
|
deployment := &apps.Deployment{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: instanceName(nfs, cfg.ID),
|
|
Namespace: nfs.Namespace,
|
|
Labels: getLabels(nfs, cfg.ID),
|
|
OwnerReferences: ownerRefs(nfs),
|
|
},
|
|
}
|
|
k8sutil.AddRookVersionLabelToDeployment(deployment)
|
|
controller.AddCephVersionLabelToDeployment(c.clusterInfo.CephVersion, deployment)
|
|
nfs.Spec.Server.Annotations.ApplyToObjectMeta(&deployment.ObjectMeta)
|
|
|
|
cephConfigVol, _ := cephConfigVolumeAndMount()
|
|
nfsConfigVol, _ := nfsConfigVolumeAndMount(cfg.ConfigConfigMap)
|
|
dbusVol, _ := dbusVolumeAndMount()
|
|
podSpec := v1.PodSpec{
|
|
InitContainers: []v1.Container{
|
|
c.connectionConfigInitContainer(nfs),
|
|
},
|
|
Containers: []v1.Container{
|
|
c.daemonContainer(nfs, cfg),
|
|
c.dbusContainer(nfs), // dbus sidecar
|
|
},
|
|
RestartPolicy: v1.RestartPolicyAlways,
|
|
Volumes: []v1.Volume{
|
|
// do not mount usual daemon volumes, as no data is stored for this daemon, and the ceph
|
|
// config file is generated by the init container. we don't need to worry about missing
|
|
// override configs, because nfs-ganesha is not a Ceph daemon; it wouldn't observe any
|
|
// overrides anyway
|
|
cephConfigVol,
|
|
keyring.Volume().Admin(),
|
|
nfsConfigVol,
|
|
dbusVol,
|
|
},
|
|
HostNetwork: c.clusterSpec.Network.IsHost(),
|
|
PriorityClassName: nfs.Spec.Server.PriorityClassName,
|
|
}
|
|
// Replace default unreachable node toleration
|
|
k8sutil.AddUnreachableNodeToleration(&podSpec)
|
|
|
|
if c.clusterSpec.Network.IsHost() {
|
|
podSpec.DNSPolicy = v1.DNSClusterFirstWithHostNet
|
|
}
|
|
nfs.Spec.Server.Placement.ApplyToPodSpec(&podSpec)
|
|
|
|
podTemplateSpec := v1.PodTemplateSpec{
|
|
ObjectMeta: metav1.ObjectMeta{
|
|
Name: instanceName(nfs, cfg.ID),
|
|
Labels: getLabels(nfs, cfg.ID),
|
|
},
|
|
Spec: podSpec,
|
|
}
|
|
nfs.Spec.Server.Annotations.ApplyToObjectMeta(&podTemplateSpec.ObjectMeta)
|
|
|
|
// Multiple replicas of the nfs service would be handled by creating a service and a new deployment for each one, rather than increasing the pod count here
|
|
replicas := int32(1)
|
|
deployment.Spec = apps.DeploymentSpec{
|
|
Selector: &metav1.LabelSelector{
|
|
MatchLabels: getLabels(nfs, cfg.ID),
|
|
},
|
|
Template: podTemplateSpec,
|
|
Replicas: &replicas,
|
|
}
|
|
return deployment
|
|
}
|
|
|
|
func (c *CephNFSController) connectionConfigInitContainer(nfs cephv1.CephNFS) v1.Container {
|
|
_, cephConfigMount := cephConfigVolumeAndMount()
|
|
|
|
return controller.GenerateMinimalCephConfInitContainer(
|
|
"client.admin",
|
|
keyring.VolumeMount().AdminKeyringFilePath(),
|
|
c.clusterSpec.CephVersion.Image,
|
|
[]v1.VolumeMount{
|
|
cephConfigMount,
|
|
keyring.VolumeMount().Admin(),
|
|
},
|
|
nfs.Spec.Server.Resources,
|
|
mon.PodSecurityContext(),
|
|
)
|
|
}
|
|
|
|
func (c *CephNFSController) daemonContainer(nfs cephv1.CephNFS, cfg daemonConfig) v1.Container {
|
|
_, cephConfigMount := cephConfigVolumeAndMount()
|
|
_, nfsConfigMount := nfsConfigVolumeAndMount(cfg.ConfigConfigMap)
|
|
_, dbusMount := dbusVolumeAndMount()
|
|
|
|
return v1.Container{
|
|
Name: "nfs-ganesha",
|
|
Command: []string{
|
|
"ganesha.nfsd",
|
|
},
|
|
Args: []string{
|
|
"-F", // foreground
|
|
"-L", "STDERR", // log to stderr
|
|
"-p", ganeshaPid, // PID file location
|
|
},
|
|
Image: c.clusterSpec.CephVersion.Image,
|
|
VolumeMounts: []v1.VolumeMount{
|
|
cephConfigMount,
|
|
keyring.VolumeMount().Admin(),
|
|
nfsConfigMount,
|
|
dbusMount,
|
|
},
|
|
Env: append(
|
|
controller.DaemonEnvVars(c.clusterSpec.CephVersion.Image),
|
|
),
|
|
Resources: nfs.Spec.Server.Resources,
|
|
SecurityContext: mon.PodSecurityContext(),
|
|
}
|
|
}
|
|
|
|
func (c *CephNFSController) dbusContainer(nfs cephv1.CephNFS) v1.Container {
|
|
_, dbusMount := dbusVolumeAndMount()
|
|
return v1.Container{
|
|
Name: "dbus-daemon",
|
|
Command: []string{
|
|
"dbus-daemon",
|
|
},
|
|
Args: []string{
|
|
"--nofork", // run in foreground
|
|
"--system", // use system config file (uses /run/dbus/system_bus_socket)
|
|
"--nopidfile", // don't write a pid file
|
|
// some dbus-daemon versions have flag --nosyslog to send logs to sterr; not ceph upstream image
|
|
},
|
|
Image: c.clusterSpec.CephVersion.Image,
|
|
VolumeMounts: []v1.VolumeMount{
|
|
dbusMount,
|
|
},
|
|
Env: append(
|
|
// do not need access to Ceph env vars b/c not a Ceph daemon
|
|
k8sutil.ClusterDaemonEnvVars(c.clusterSpec.CephVersion.Image),
|
|
),
|
|
Resources: nfs.Spec.Server.Resources,
|
|
}
|
|
}
|
|
|
|
func getLabels(n cephv1.CephNFS, name string) map[string]string {
|
|
labels := controller.AppLabels(appName, n.Namespace)
|
|
labels["ceph_nfs"] = n.Name
|
|
labels["instance"] = name
|
|
return labels
|
|
}
|
|
|
|
func cephConfigVolumeAndMount() (v1.Volume, v1.VolumeMount) {
|
|
// nfs ganesha produces its own ceph config file, so cannot use controller.DaemonVolume or
|
|
// controller.DaemonVolumeMounts since that will bring in global ceph config file
|
|
cfgDir := cephconfig.DefaultConfigDir
|
|
volName := k8sutil.PathToVolumeName(cfgDir)
|
|
v := v1.Volume{Name: volName, VolumeSource: v1.VolumeSource{EmptyDir: &v1.EmptyDirVolumeSource{}}}
|
|
m := v1.VolumeMount{Name: volName, MountPath: cfgDir}
|
|
return v, m
|
|
}
|
|
|
|
func nfsConfigVolumeAndMount(configConfigMap string) (v1.Volume, v1.VolumeMount) {
|
|
cfgDir := "/etc/ganesha" // cfg file: /etc/ganesha/ganesha.conf
|
|
cfgVolName := ganeshaConfigVolume
|
|
configMapSource := &v1.ConfigMapVolumeSource{
|
|
LocalObjectReference: v1.LocalObjectReference{Name: configConfigMap},
|
|
Items: []v1.KeyToPath{{Key: "config", Path: "ganesha.conf"}},
|
|
}
|
|
v := v1.Volume{Name: cfgVolName, VolumeSource: v1.VolumeSource{ConfigMap: configMapSource}}
|
|
m := v1.VolumeMount{Name: cfgVolName, MountPath: cfgDir}
|
|
return v, m
|
|
}
|
|
|
|
func dbusVolumeAndMount() (v1.Volume, v1.VolumeMount) {
|
|
dbusSocketDir := "/run/dbus" // socket file: /run/dbus/system_bus_socket
|
|
volName := k8sutil.PathToVolumeName(dbusSocketDir)
|
|
v := v1.Volume{Name: volName, VolumeSource: v1.VolumeSource{EmptyDir: &v1.EmptyDirVolumeSource{}}}
|
|
m := v1.VolumeMount{Name: volName, MountPath: dbusSocketDir}
|
|
return v, m
|
|
}
|