We now always perform ok-to-stop change when the deployment is going to be updated. This could be for various reasons like change in the spec or a new Ceph image. In any case, if Ceph resources are going to be restarted this must be done carefully. Also trimmed the isUpgrade variable, the upcoming commit will remove the need of that variable. Keeping isUpgrade from the controller code since we need it to detect whether an orchestration upgrade should be triggered or not (based on the ceph cluster status). Also, we keep it on the child controllers so we don't go into an orchestration phase if the cluster image hasn't changed. The child does not need to be updated/go into its orchestration phase because the cluster CR changed. Closes: https://github.com/rook/rook/issues/4379 Signed-off-by: Sébastien Han <seb@redhat.com>
279 lines
9.7 KiB
Go
279 lines
9.7 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 file
|
|
|
|
import (
|
|
"fmt"
|
|
|
|
"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"
|
|
cephconfig "github.com/rook/rook/pkg/daemon/ceph/config"
|
|
"github.com/rook/rook/pkg/daemon/ceph/model"
|
|
"github.com/rook/rook/pkg/operator/ceph/file/mds"
|
|
"github.com/rook/rook/pkg/operator/ceph/pool"
|
|
cephver "github.com/rook/rook/pkg/operator/ceph/version"
|
|
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
|
|
)
|
|
|
|
const (
|
|
dataPoolSuffix = "data"
|
|
metadataPoolSuffix = "metadata"
|
|
appName = "cephfs"
|
|
)
|
|
|
|
// Filesystem represents an instance of a Ceph filesystem (CephFS)
|
|
type Filesystem struct {
|
|
Name string
|
|
metadataPool *model.Pool
|
|
dataPools []*model.Pool
|
|
activeMDSCount int32
|
|
}
|
|
|
|
// createFilesystem creates a Ceph filesystem with metadata servers
|
|
func createFilesystem(
|
|
clusterInfo *cephconfig.ClusterInfo,
|
|
context *clusterd.Context,
|
|
fs cephv1.CephFilesystem,
|
|
rookVersion string,
|
|
clusterSpec *cephv1.ClusterSpec,
|
|
ownerRefs metav1.OwnerReference,
|
|
dataDirHostPath string,
|
|
) error {
|
|
if err := validateFilesystem(context, fs); err != nil {
|
|
return err
|
|
}
|
|
|
|
if len(fs.Spec.DataPools) != 0 {
|
|
var dataPools []*model.Pool
|
|
for _, p := range fs.Spec.DataPools {
|
|
dataPools = append(dataPools, p.ToModel(""))
|
|
}
|
|
f := newFS(fs.Name, fs.Spec.MetadataPool.ToModel(""), dataPools, fs.Spec.MetadataServer.ActiveCount)
|
|
if err := f.doFilesystemCreate(context, clusterInfo.CephVersion, fs.Namespace); err != nil {
|
|
return errors.Wrapf(err, "failed to create filesystem %s", fs.Name)
|
|
}
|
|
}
|
|
|
|
filesystem, err := client.GetFilesystem(context, fs.Namespace, fs.Name)
|
|
if err != nil {
|
|
return errors.Wrapf(err, "failed to get filesystem %s", fs.Name)
|
|
}
|
|
|
|
// As of Nautilus, allow_standby_replay is a fs property so we need to apply it
|
|
if clusterInfo.CephVersion.IsAtLeastNautilus() {
|
|
if fs.Spec.MetadataServer.ActiveStandby {
|
|
if err = client.AllowStandbyReplay(context, fs.Namespace, fs.Name, fs.Spec.MetadataServer.ActiveStandby); err != nil {
|
|
return errors.Wrapf(err, "failed to set allow_standby_replay to filesystem %s", fs.Name)
|
|
}
|
|
}
|
|
}
|
|
|
|
// set the number of active mds instances
|
|
if fs.Spec.MetadataServer.ActiveCount > 1 {
|
|
if err = client.SetNumMDSRanks(context, clusterInfo.CephVersion, fs.Namespace, fs.Name, fs.Spec.MetadataServer.ActiveCount); err != nil {
|
|
logger.Warningf("failed setting active mds count to %d. %v", fs.Spec.MetadataServer.ActiveCount, err)
|
|
}
|
|
}
|
|
|
|
logger.Infof("start running mdses for filesystem %s", fs.Name)
|
|
c := mds.NewCluster(clusterInfo, context, rookVersion, clusterSpec, fs, filesystem, ownerRefs, dataDirHostPath)
|
|
if err := c.Start(); err != nil {
|
|
return err
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// deleteFileSystem deletes the filesystem from Ceph
|
|
func deleteFilesystem(context *clusterd.Context, cephVersion cephver.CephVersion, fs cephv1.CephFilesystem) error {
|
|
// The most important part of deletion is that the filesystem gets removed from Ceph
|
|
// The K8s resources will already be removed with the K8s owner references
|
|
if err := downFilesystem(context, cephVersion, fs.Namespace, fs.Name); err != nil {
|
|
// If the fs isn't deleted from Ceph, leave the daemons so it can still be used.
|
|
return errors.Wrapf(err, "failed to down filesystem %q", fs.Name)
|
|
}
|
|
|
|
// Permanently remove the filesystem if it was created by rook
|
|
if len(fs.Spec.DataPools) != 0 {
|
|
if err := client.RemoveFilesystem(context, fs.Namespace, fs.Name, fs.Spec.PreservePoolsOnDelete); err != nil {
|
|
return errors.Wrapf(err, "failed to remove filesystem %q", fs.Name)
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
func validateFilesystem(context *clusterd.Context, f cephv1.CephFilesystem) error {
|
|
if f.Name == "" {
|
|
return errors.New("missing name")
|
|
}
|
|
if f.Namespace == "" {
|
|
return errors.New("missing namespace")
|
|
}
|
|
if f.Spec.MetadataServer.ActiveCount < 1 {
|
|
return errors.New("MetadataServer.ActiveCount must be at least 1")
|
|
}
|
|
// No data pool means that we expect the fs to exist already
|
|
if len(f.Spec.DataPools) == 0 {
|
|
return nil
|
|
}
|
|
if err := pool.ValidatePoolSpec(context, f.Namespace, &f.Spec.MetadataPool); err != nil {
|
|
return errors.Wrapf(err, "invalid metadata pool")
|
|
}
|
|
for _, p := range f.Spec.DataPools {
|
|
if err := pool.ValidatePoolSpec(context, f.Namespace, &p); err != nil {
|
|
return errors.Wrapf(err, "Invalid data pool")
|
|
}
|
|
}
|
|
|
|
return nil
|
|
}
|
|
|
|
// newFS creates a new instance of the file (MDS) service
|
|
func newFS(name string, metadataPool *model.Pool, dataPools []*model.Pool, activeMDSCount int32) *Filesystem {
|
|
|
|
metadataPool.Name = fmt.Sprintf("%s-%s", name, metadataPoolSuffix)
|
|
for i, pool := range dataPools {
|
|
pool.Name = fmt.Sprintf("%s-%s%d", name, dataPoolSuffix, i)
|
|
}
|
|
|
|
// For the filesystem pool we don't want to enable the application pool
|
|
// since it's being done via 'fs new' already
|
|
metadataPool.NotEnableAppPool = true
|
|
|
|
return &Filesystem{
|
|
Name: name,
|
|
metadataPool: metadataPool,
|
|
dataPools: dataPools,
|
|
activeMDSCount: activeMDSCount,
|
|
}
|
|
}
|
|
|
|
// doFilesystemCreate starts the Ceph file daemons and creates the filesystem in Ceph.
|
|
func (f *Filesystem) doFilesystemCreate(context *clusterd.Context, cephVersion cephver.CephVersion, clusterName string) error {
|
|
_, err := client.GetFilesystem(context, clusterName, f.Name)
|
|
if err == nil {
|
|
logger.Infof("filesystem %s already exists", f.Name)
|
|
// Even if the fs already exists, the num active mdses may have changed
|
|
|
|
if err := client.SetNumMDSRanks(context, cephVersion, clusterName, f.Name, f.activeMDSCount); err != nil {
|
|
logger.Errorf(
|
|
fmt.Sprintf("failed to set num mds ranks (max_mds) to %d for filesystem %s, still continuing. ", f.activeMDSCount, f.Name) +
|
|
"this error is not critical, but mdses may not be as failure tolerant as desired. " +
|
|
fmt.Sprintf("USER should verify that the number of active mdses is %d with 'ceph fs get %s'", f.activeMDSCount, f.Name) +
|
|
fmt.Sprintf(". %v", err),
|
|
)
|
|
}
|
|
return nil
|
|
}
|
|
if len(f.dataPools) == 0 {
|
|
return errors.New("at least one data pool must be specified")
|
|
}
|
|
|
|
fslist, err := client.ListFilesystems(context, clusterName)
|
|
if err != nil {
|
|
return errors.Wrapf(err, "Unable to list existing filesystem")
|
|
}
|
|
if len(fslist) > 0 && !client.IsMultiFSEnabled() {
|
|
return errors.Errorf("cannot create multiple filesystems. enable %s env variable to create more than one", client.MultiFsEnv)
|
|
}
|
|
|
|
poolNames, err := client.GetPoolNamesByID(context, clusterName)
|
|
if err != nil {
|
|
return errors.Wrapf(err, "failed to get pool names")
|
|
}
|
|
|
|
logger.Infof("Creating filesystem %s", f.Name)
|
|
|
|
// Make easy to locate a pool by name and avoid repeated searches
|
|
reversedPoolMap := make(map[string]int)
|
|
for key, value := range poolNames {
|
|
reversedPoolMap[value] = key
|
|
}
|
|
|
|
poolsCreated := false
|
|
if _, poolFound := reversedPoolMap[f.metadataPool.Name]; !poolFound {
|
|
poolsCreated = true
|
|
err = client.CreatePoolWithProfile(context, clusterName, *f.metadataPool, appName)
|
|
if err != nil {
|
|
return errors.Wrapf(err, "failed to create metadata pool %q", f.metadataPool.Name)
|
|
}
|
|
}
|
|
|
|
var dataPoolNames []string
|
|
for _, pool := range f.dataPools {
|
|
dataPoolNames = append(dataPoolNames, pool.Name)
|
|
if _, poolFound := reversedPoolMap[pool.Name]; !poolFound {
|
|
poolsCreated = true
|
|
err = client.CreatePoolWithProfile(context, clusterName, *pool, appName)
|
|
if err != nil {
|
|
return errors.Wrapf(err, "failed to create data pool %q", pool.Name)
|
|
}
|
|
if pool.Type == model.ErasureCoded {
|
|
// An erasure coded data pool used for a filesystem must allow overwrites
|
|
if err := client.SetPoolProperty(context, clusterName, pool.Name, "allow_ec_overwrites", "true"); err != nil {
|
|
logger.Warningf("failed to set ec pool property. %v", err)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// create the filesystem ('fs new' needs to be forced in order to reuse pre-existing pools)
|
|
// if only one pool is created new it wont work (to avoid inconsistencies).
|
|
if err := client.CreateFilesystem(context, clusterName, f.Name, f.metadataPool.Name, dataPoolNames, !poolsCreated); err != nil {
|
|
return err
|
|
}
|
|
|
|
logger.Infof("created filesystem %s on %d data pool(s) and metadata pool %s", f.Name, len(f.dataPools), f.metadataPool.Name)
|
|
return nil
|
|
}
|
|
|
|
// downFilesystem marks the filesystem as down and the MDS' as failed
|
|
func downFilesystem(context *clusterd.Context, cephVersion cephver.CephVersion, clusterName, filesystemName string) error {
|
|
logger.Infof("Downing filesystem %s", filesystemName)
|
|
|
|
// From Ceph nautilus onwards, a single Ceph command marks the filesystem as down and
|
|
// MDSes as failed
|
|
if cephVersion.IsAtLeastNautilus() {
|
|
if err := client.FailFilesystem(context, clusterName, filesystemName); err != nil {
|
|
return err
|
|
}
|
|
logger.Infof("Downed filesystem %s", filesystemName)
|
|
return nil
|
|
}
|
|
|
|
// mark the cephFS instance as cluster_down before removing
|
|
if err := client.MarkFilesystemAsDown(context, clusterName, filesystemName); err != nil {
|
|
return err
|
|
}
|
|
|
|
// mark each MDS associated with the filesystem to "failed"
|
|
fsDetails, err := client.GetFilesystem(context, clusterName, filesystemName)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
for _, mdsInfo := range fsDetails.MDSMap.Info {
|
|
if err := client.FailMDS(context, clusterName, mdsInfo.GID); err != nil {
|
|
return err
|
|
}
|
|
}
|
|
|
|
logger.Infof("Downed filesystem %s", filesystemName)
|
|
return nil
|
|
}
|