Blaine Gardner
573a47e736
When IPv6-only network mode is enabled and when mons are configured to listen only on msgrV2 port, Rook has to add the port to the mon --public-bind-addr in order to force mons to stop listening on the msgrV1 port. This can cause problems for IPv6 clusters where adding ':<port>' to the end of an IP addr looks like a valid IPv6 address. In IPv6 mode, Rook needs to set this to '[<ip>]:<port>' with the brackets to be a non-ambiguous IPv6 address. In IPv4 mode, Rook needs to keep its existing behavior. IPv4 mode should also continue to be the defualt assumption if the network config isn't specified. If 'dualStack: true' is specified, Rook is informed that the environment is dual stack, but Rook doesn't know which IP family to expect from the pod IP. In this scenario, it is safest for Rook not to apply the port, since assuming either IPv4 or IPv6 format could be incorrect. In this scenario, mons may still listen on msgrV1 ports, but this is preferable to an environment where mons perpetually crash. Signed-off-by: Blaine Gardner <blaine.gardner@ibm.com>
417 lines
15 KiB
Go
417 lines
15 KiB
Go
/*
|
||
Copyright 2016 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 mon
|
||
|
||
import (
|
||
"fmt"
|
||
"path"
|
||
"strings"
|
||
|
||
"github.com/pkg/errors"
|
||
cephv1 "github.com/rook/rook/pkg/apis/ceph.rook.io/v1"
|
||
"github.com/rook/rook/pkg/clusterd"
|
||
"github.com/rook/rook/pkg/daemon/ceph/client"
|
||
"github.com/rook/rook/pkg/operator/ceph/config"
|
||
"github.com/rook/rook/pkg/operator/ceph/controller"
|
||
"github.com/rook/rook/pkg/operator/k8sutil"
|
||
apps "k8s.io/api/apps/v1"
|
||
corev1 "k8s.io/api/core/v1"
|
||
"k8s.io/apimachinery/pkg/api/resource"
|
||
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
|
||
)
|
||
|
||
const (
|
||
// Full path of the command used to invoke the Ceph mon daemon
|
||
cephMonCommand = "ceph-mon"
|
||
)
|
||
|
||
func (c *Cluster) getLabels(monConfig *monConfig, canary, includeNewLabels bool) map[string]string {
|
||
// Mons have a service for each mon, so the additional pod data is relevant for its services
|
||
// Use pod labels to keep "mon: id" for legacy
|
||
labels := controller.CephDaemonAppLabels(AppName, c.Namespace, config.MonType, monConfig.DaemonName, c.ClusterInfo.NamespacedName().Name, "cephclusters.ceph.rook.io", includeNewLabels)
|
||
// Add "mon_cluster: <namespace>" for legacy
|
||
labels[monClusterAttr] = c.Namespace
|
||
if canary {
|
||
labels["mon_canary"] = "true"
|
||
}
|
||
if includeNewLabels {
|
||
monVolumeClaimTemplate := c.monVolumeClaimTemplate(monConfig)
|
||
if monVolumeClaimTemplate != nil {
|
||
size := monVolumeClaimTemplate.Spec.Resources.Requests[corev1.ResourceStorage]
|
||
labels["pvc_name"] = monConfig.ResourceName
|
||
labels["pvc_size"] = size.String()
|
||
}
|
||
if monConfig.Zone != "" {
|
||
labels["zone"] = monConfig.Zone
|
||
}
|
||
}
|
||
|
||
return labels
|
||
}
|
||
|
||
func (c *Cluster) getFailureDomainName() string {
|
||
label := GetFailureDomainLabel(c.spec)
|
||
index := strings.Index(label, "/")
|
||
if index == -1 {
|
||
return label
|
||
}
|
||
return label[index+1:]
|
||
}
|
||
|
||
func GetFailureDomainLabel(spec cephv1.ClusterSpec) string {
|
||
|
||
if spec.IsStretchCluster() && spec.Mon.StretchCluster.FailureDomainLabel != "" {
|
||
return spec.Mon.StretchCluster.FailureDomainLabel
|
||
}
|
||
|
||
if spec.ZonesRequired() && spec.Mon.FailureDomainLabel != "" {
|
||
return spec.Mon.StretchCluster.FailureDomainLabel
|
||
}
|
||
// The default topology label is for a zone
|
||
return corev1.LabelZoneFailureDomainStable
|
||
}
|
||
|
||
func (c *Cluster) makeDeployment(monConfig *monConfig, canary bool) (*apps.Deployment, error) {
|
||
d := &apps.Deployment{
|
||
ObjectMeta: metav1.ObjectMeta{
|
||
Name: monConfig.ResourceName,
|
||
Namespace: c.Namespace,
|
||
Labels: c.getLabels(monConfig, canary, true),
|
||
},
|
||
}
|
||
k8sutil.AddRookVersionLabelToDeployment(d)
|
||
cephv1.GetMonAnnotations(c.spec.Annotations).ApplyToObjectMeta(&d.ObjectMeta)
|
||
cephv1.GetMonLabels(c.spec.Labels).ApplyToObjectMeta(&d.ObjectMeta)
|
||
controller.AddCephVersionLabelToDeployment(c.ClusterInfo.CephVersion, d)
|
||
err := c.ownerInfo.SetControllerReference(d)
|
||
if err != nil {
|
||
return nil, errors.Wrapf(err, "failed to set owner reference to mon deployment %q", d.Name)
|
||
}
|
||
|
||
pod, err := c.makeMonPod(monConfig, canary)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
replicaCount := int32(1)
|
||
d.Spec = apps.DeploymentSpec{
|
||
Selector: &metav1.LabelSelector{
|
||
MatchLabels: c.getLabels(monConfig, canary, false),
|
||
},
|
||
Template: corev1.PodTemplateSpec{
|
||
ObjectMeta: pod.ObjectMeta,
|
||
Spec: pod.Spec,
|
||
},
|
||
Replicas: &replicaCount,
|
||
Strategy: apps.DeploymentStrategy{
|
||
Type: apps.RecreateDeploymentStrategyType,
|
||
},
|
||
}
|
||
|
||
return d, nil
|
||
}
|
||
|
||
func (c *Cluster) makeDeploymentPVC(m *monConfig, canary bool) (*corev1.PersistentVolumeClaim, error) {
|
||
template := c.monVolumeClaimTemplate(m)
|
||
volumeMode := corev1.PersistentVolumeFilesystem
|
||
pvc := &corev1.PersistentVolumeClaim{
|
||
ObjectMeta: metav1.ObjectMeta{
|
||
Name: m.ResourceName,
|
||
Namespace: c.Namespace,
|
||
Labels: c.getLabels(m, canary, true),
|
||
},
|
||
Spec: corev1.PersistentVolumeClaimSpec{
|
||
AccessModes: []corev1.PersistentVolumeAccessMode{
|
||
corev1.ReadWriteOnce,
|
||
},
|
||
Resources: template.Spec.Resources,
|
||
StorageClassName: template.Spec.StorageClassName,
|
||
VolumeMode: &volumeMode,
|
||
},
|
||
}
|
||
k8sutil.AddRookVersionLabelToObjectMeta(&pvc.ObjectMeta)
|
||
cephv1.GetMonAnnotations(c.spec.Annotations).ApplyToObjectMeta(&pvc.ObjectMeta)
|
||
controller.AddCephVersionLabelToObjectMeta(c.ClusterInfo.CephVersion, &pvc.ObjectMeta)
|
||
err := c.ownerInfo.SetControllerReference(pvc)
|
||
if err != nil {
|
||
return nil, errors.Wrapf(err, "failed to set owner reference to mon pvc %q", pvc.Name)
|
||
}
|
||
|
||
// k8s uses limit as the resource request fallback
|
||
if _, ok := pvc.Spec.Resources.Limits[corev1.ResourceStorage]; ok {
|
||
return pvc, nil
|
||
}
|
||
|
||
// specific request in the crd
|
||
if _, ok := pvc.Spec.Resources.Requests[corev1.ResourceStorage]; ok {
|
||
return pvc, nil
|
||
}
|
||
|
||
req, err := resource.ParseQuantity(cephMonDefaultStorageRequest)
|
||
if err != nil {
|
||
return nil, err
|
||
}
|
||
|
||
if pvc.Spec.Resources.Requests == nil {
|
||
pvc.Spec.Resources.Requests = corev1.ResourceList{}
|
||
}
|
||
pvc.Spec.Resources.Requests[corev1.ResourceStorage] = req
|
||
|
||
return pvc, nil
|
||
}
|
||
|
||
func (c *Cluster) makeMonPod(monConfig *monConfig, canary bool) (*corev1.Pod, error) {
|
||
logger.Debugf("monConfig: %+v", monConfig)
|
||
podSpec := corev1.PodSpec{
|
||
InitContainers: []corev1.Container{
|
||
c.makeChownInitContainer(monConfig),
|
||
c.makeMonFSInitContainer(monConfig),
|
||
},
|
||
Containers: []corev1.Container{
|
||
c.makeMonDaemonContainer(monConfig),
|
||
},
|
||
RestartPolicy: corev1.RestartPolicyAlways,
|
||
// we decide later whether to use a PVC volume or host volumes for mons, so only populate
|
||
// the base volumes at this point.
|
||
Volumes: controller.DaemonVolumesBase(monConfig.DataPathMap, keyringStoreName, c.spec.DataDirHostPath),
|
||
HostNetwork: monConfig.UseHostNetwork,
|
||
PriorityClassName: cephv1.GetMonPriorityClassName(c.spec.PriorityClassNames),
|
||
ServiceAccountName: k8sutil.DefaultServiceAccount,
|
||
}
|
||
|
||
// If the log collector is enabled we add the side-car container
|
||
if c.spec.LogCollector.Enabled {
|
||
shareProcessNamespace := true
|
||
podSpec.ShareProcessNamespace = &shareProcessNamespace
|
||
podSpec.Containers = append(podSpec.Containers, *controller.LogCollectorContainer(fmt.Sprintf("%s.%s", cephMonCommand, monConfig.DaemonName), c.ClusterInfo.Namespace, c.spec))
|
||
}
|
||
|
||
// Replace default unreachable node toleration
|
||
if c.monVolumeClaimTemplate(monConfig) != nil {
|
||
k8sutil.AddUnreachableNodeToleration(&podSpec)
|
||
}
|
||
|
||
pod := &corev1.Pod{
|
||
ObjectMeta: metav1.ObjectMeta{
|
||
Name: monConfig.ResourceName,
|
||
Namespace: c.Namespace,
|
||
Labels: c.getLabels(monConfig, canary, true),
|
||
},
|
||
Spec: podSpec,
|
||
}
|
||
cephv1.GetMonAnnotations(c.spec.Annotations).ApplyToObjectMeta(&pod.ObjectMeta)
|
||
cephv1.GetMonLabels(c.spec.Labels).ApplyToObjectMeta(&pod.ObjectMeta)
|
||
|
||
if monConfig.UseHostNetwork {
|
||
pod.Spec.DNSPolicy = corev1.DNSClusterFirstWithHostNet
|
||
} else if c.spec.Network.IsMultus() {
|
||
cluster := cephv1.CephCluster{
|
||
ObjectMeta: metav1.ObjectMeta{
|
||
Namespace: c.Namespace,
|
||
},
|
||
Spec: c.spec,
|
||
}
|
||
if err := k8sutil.ApplyMultus(cluster.GetNamespace(), &cluster.Spec.Network, &pod.ObjectMeta); err != nil {
|
||
return nil, err
|
||
}
|
||
}
|
||
|
||
if c.spec.ZonesRequired() {
|
||
nodeAffinity, err := k8sutil.GenerateNodeAffinity(fmt.Sprintf("%s=%s", GetFailureDomainLabel(c.spec), monConfig.Zone))
|
||
if err != nil {
|
||
return nil, errors.Wrapf(err, "failed to generate mon %q node affinity", monConfig.DaemonName)
|
||
}
|
||
pod.Spec.Affinity = &corev1.Affinity{NodeAffinity: nodeAffinity}
|
||
}
|
||
|
||
return pod, nil
|
||
}
|
||
|
||
/*
|
||
* Container specs
|
||
*/
|
||
|
||
// Init and daemon containers require the same context, so we call it 'pod' context
|
||
|
||
func (c *Cluster) makeChownInitContainer(monConfig *monConfig) corev1.Container {
|
||
return controller.ChownCephDataDirsInitContainer(
|
||
*monConfig.DataPathMap,
|
||
c.spec.CephVersion.Image,
|
||
controller.GetContainerImagePullPolicy(c.spec.CephVersion.ImagePullPolicy),
|
||
controller.DaemonVolumeMounts(monConfig.DataPathMap, keyringStoreName, c.spec.DataDirHostPath),
|
||
cephv1.GetMonResources(c.spec.Resources),
|
||
controller.PodSecurityContext(),
|
||
"",
|
||
)
|
||
}
|
||
|
||
func (c *Cluster) makeMonFSInitContainer(monConfig *monConfig) corev1.Container {
|
||
return corev1.Container{
|
||
Name: "init-mon-fs",
|
||
Command: []string{
|
||
cephMonCommand,
|
||
},
|
||
Args: append(
|
||
controller.DaemonFlags(c.ClusterInfo, &c.spec, monConfig.DaemonName),
|
||
// needed so we can generate an initial monmap
|
||
// otherwise the mkfs will say: "0 no local addrs match monmap"
|
||
config.NewFlag("public-addr", monConfig.PublicIP),
|
||
"--mkfs",
|
||
),
|
||
Image: c.spec.CephVersion.Image,
|
||
ImagePullPolicy: controller.GetContainerImagePullPolicy(c.spec.CephVersion.ImagePullPolicy),
|
||
VolumeMounts: controller.DaemonVolumeMounts(monConfig.DataPathMap, keyringStoreName, c.spec.DataDirHostPath),
|
||
SecurityContext: controller.PodSecurityContext(),
|
||
// filesystem creation does not require ports to be exposed
|
||
Env: controller.DaemonEnvVars(&c.spec),
|
||
Resources: cephv1.GetMonResources(c.spec.Resources),
|
||
}
|
||
}
|
||
|
||
func (c *Cluster) makeMonDaemonContainer(monConfig *monConfig) corev1.Container {
|
||
podIPEnvVar := "ROOK_POD_IP"
|
||
|
||
container := corev1.Container{
|
||
Name: "mon",
|
||
Command: []string{
|
||
cephMonCommand,
|
||
},
|
||
Args: append(
|
||
controller.DaemonFlags(c.ClusterInfo, &c.spec, monConfig.DaemonName),
|
||
"--foreground",
|
||
// If the mon is already in the monmap, when the port is left off of --public-addr,
|
||
// it will still advertise on the previous port b/c monmap is saved to mon database.
|
||
config.NewFlag("public-addr", monConfig.PublicIP),
|
||
// Set '--setuser-match-path' so that existing directory owned by root won't affect the daemon startup.
|
||
// For existing data store owned by root, the daemon will continue to run as root
|
||
//
|
||
// We use 'store.db' here because during an upgrade the init container will set 'ceph:ceph' to monConfig.DataPathMap.ContainerDataDir
|
||
// but inside the permissions will be 'root:root' AND we don't want to chown recursively on the mon data directory
|
||
// We want to avoid potential startup time issue if the store is big
|
||
config.NewFlag("setuser-match-path", path.Join(monConfig.DataPathMap.ContainerDataDir, "store.db")),
|
||
),
|
||
Image: c.spec.CephVersion.Image,
|
||
ImagePullPolicy: controller.GetContainerImagePullPolicy(c.spec.CephVersion.ImagePullPolicy),
|
||
VolumeMounts: controller.DaemonVolumeMounts(monConfig.DataPathMap, keyringStoreName, c.spec.DataDirHostPath),
|
||
SecurityContext: controller.PodSecurityContext(),
|
||
Ports: []corev1.ContainerPort{
|
||
{
|
||
Name: "tcp-msgr2",
|
||
ContainerPort: DefaultMsgr2Port,
|
||
Protocol: corev1.ProtocolTCP,
|
||
},
|
||
},
|
||
Env: append(
|
||
controller.DaemonEnvVars(&c.spec),
|
||
k8sutil.PodIPEnvVar(podIPEnvVar),
|
||
),
|
||
Resources: cephv1.GetMonResources(c.spec.Resources),
|
||
StartupProbe: controller.GenerateStartupProbeExecDaemon(config.MonType, monConfig.DaemonName),
|
||
LivenessProbe: controller.GenerateLivenessProbeExecDaemon(config.MonType, monConfig.DaemonName),
|
||
WorkingDir: config.VarLogCephDir,
|
||
}
|
||
|
||
bindaddr := controller.ContainerEnvVarReference(podIPEnvVar)
|
||
if monConfig.Port == DefaultMsgr2Port {
|
||
container.Args = append(container.Args, config.NewFlag("ms_bind_msgr1", "false"))
|
||
|
||
// mons don't use --ms-bind-msgr1 to control whether they bind to v1 port or not.
|
||
// in order to force use of only v2 port, Rook must include the port in the bind addr
|
||
if c.spec.Network.DualStack {
|
||
// in a dual stack environment, Rook can't know whether IPv4 or IPv6 will be used.
|
||
// in order to be safe, don't add the port to the bind addr. this will mean that mons
|
||
// might listen on both msgr1 and msgr2 ports, but it is more critical to make sure mons
|
||
// don't crash than to forcefully disable msgr1
|
||
} else if c.spec.Network.IPFamily == cephv1.IPv6 {
|
||
// IPv6 addrs have to be surrounded in square brackets when a port is given
|
||
bindaddr = fmt.Sprintf("[%s]:%d", bindaddr, DefaultMsgr2Port)
|
||
} else if c.spec.Network.IPFamily == cephv1.IPv4 || c.spec.Network.IPFamily == "" {
|
||
// IPv4 addrs must have the port added without any special syntax
|
||
// if the IP family is unset, IPv4 is a safe assumption
|
||
bindaddr = fmt.Sprintf("%s:%d", bindaddr, DefaultMsgr2Port)
|
||
}
|
||
} else {
|
||
// Add messenger 1 port
|
||
container.Ports = append(container.Ports, corev1.ContainerPort{
|
||
Name: "tcp-msgr1",
|
||
ContainerPort: monConfig.Port,
|
||
Protocol: corev1.ProtocolTCP,
|
||
})
|
||
}
|
||
|
||
if monConfig.Zone != "" {
|
||
desiredLocation := fmt.Sprintf("%s=%s", c.getFailureDomainName(), monConfig.Zone)
|
||
container.Args = append(container.Args, []string{"--set-crush-location", desiredLocation}...)
|
||
if monConfig.Zone == c.getArbiterZone() {
|
||
// remember the arbiter mon to be set later in the reconcile after the OSDs are configured
|
||
c.arbiterMon = monConfig.DaemonName
|
||
}
|
||
}
|
||
|
||
container = config.ConfigureStartupProbe(container, c.spec.HealthCheck.StartupProbe[cephv1.KeyMon])
|
||
container = config.ConfigureLivenessProbe(container, c.spec.HealthCheck.LivenessProbe[cephv1.KeyMon])
|
||
|
||
// If host networking is enabled, we don't need a bind addr that is different from the public addr
|
||
if !monConfig.UseHostNetwork {
|
||
// Opposite of the above, --public-bind-addr will *not* still advertise on the previous
|
||
// port, which makes sense because this is the pod IP, which changes with every new pod.
|
||
container.Args = append(container.Args, config.NewFlag("public-bind-addr", bindaddr))
|
||
}
|
||
|
||
return container
|
||
}
|
||
|
||
// UpdateCephDeploymentAndWait verifies a deployment can be stopped or continued
|
||
func UpdateCephDeploymentAndWait(context *clusterd.Context, clusterInfo *client.ClusterInfo, deployment *apps.Deployment, daemonType, daemonName string, skipUpgradeChecks, continueUpgradeAfterChecksEvenIfNotHealthy bool) error {
|
||
|
||
callback := func(action string) error {
|
||
// At this point, we are in an upgrade
|
||
if skipUpgradeChecks {
|
||
logger.Warningf("this is a Ceph upgrade, not performing upgrade checks because skipUpgradeChecks is %t", skipUpgradeChecks)
|
||
return nil
|
||
}
|
||
|
||
logger.Infof("checking if we can %s the deployment %s", action, deployment.Name)
|
||
|
||
if action == "stop" {
|
||
err := client.OkToStop(context, clusterInfo, deployment.Name, daemonType, daemonName)
|
||
if err != nil {
|
||
if continueUpgradeAfterChecksEvenIfNotHealthy {
|
||
logger.Infof("The %s daemon %s is not ok-to-stop but 'continueUpgradeAfterChecksEvenIfNotHealthy' is true, so proceeding to stop...", daemonType, daemonName)
|
||
return nil
|
||
}
|
||
return errors.Wrapf(err, "failed to check if we can %s the deployment %s", action, deployment.Name)
|
||
}
|
||
}
|
||
|
||
if action == "continue" {
|
||
err := client.OkToContinue(context, clusterInfo, deployment.Name, daemonType, daemonName)
|
||
if err != nil {
|
||
if continueUpgradeAfterChecksEvenIfNotHealthy {
|
||
logger.Infof("The %s daemon %s is not ok-to-continue but 'continueUpgradeAfterChecksEvenIfNotHealthy' is true, so continuing...", daemonType, daemonName)
|
||
return nil
|
||
}
|
||
return errors.Wrapf(err, "failed to check if we can %s the deployment %s", action, deployment.Name)
|
||
}
|
||
}
|
||
|
||
return nil
|
||
}
|
||
|
||
err := k8sutil.UpdateDeploymentAndWait(clusterInfo.Context, context, deployment, clusterInfo.Namespace, callback)
|
||
return err
|
||
}
|