Merge pull request #14055 from rook/mergify/bp/release-1.14/pr-14036

subvolumegroup: add support for size and datapool (backport #14036)

Documentation/CRDs/Shared-Filesystem/ceph-fs-subvolumegroup-crd.md

@@ -32,6 +32,10 @@ spec:
     distributed: 1            # distributed=<0, 1> (disabled=0)
     # export:                 # export=<0-256> (disabled=-1)
     # random:                 # random=[0.0, 1.0](disabled=0.0)
+  # Quota size of the subvolume group.
+  #quota: 10G
+  # data pool name for the subvolume group layout instead of the default data pool.
+  #dataPoolName: myfs-replicated
 ```
 
 ## Settings
@@ -48,7 +52,11 @@ If any setting is unspecified, a suitable default will be used automatically.
 
 * `filesystemName`: The metadata name of the CephFilesystem CR where the subvolume group will be created.
 
-* `pinning`: To distribute load across MDS ranks in predictable and stable ways. Reference: https://docs.ceph.com/en/latest/cephfs/fs-volumes/#pinning-subvolumes-and-subvolume-groups.
+* `quota`: Quota size of the Ceph Filesystem subvolume group.
+
+* `dataPoolName`: The data pool name for the subvolume group layout instead of the default data pool.
+
+* `pinning`: To distribute load across MDS ranks in predictable and stable ways. See the Ceph doc for [Pinning subvolume groups](https://docs.ceph.com/en/latest/cephfs/fs-volumes/#pinning-subvolumes-and-subvolume-groups).
     * `distributed`: Range: <0, 1>, for disabling it set to 0
     * `export`: Range: <0-256>, for disabling it set to -1
     * `random`: Range: [0.0, 1.0], for disabling it set to 0.0

Documentation/CRDs/specification.md

@@ -1576,6 +1576,30 @@ reference <a href="https://docs.ceph.com/en/latest/cephfs/fs-volumes/#pinning-su
 only one out of (export, distributed, random) can be set at a time</p>
 </td>
 </tr>
+<tr>
+<td>
+<code>quota</code><br/>
+<em>
+k8s.io/apimachinery/pkg/api/resource.Quantity
+</em>
+</td>
+<td>
+<em>(Optional)</em>
+<p>Quota size of the Ceph Filesystem subvolume group.</p>
+</td>
+</tr>
+<tr>
+<td>
+<code>dataPoolName</code><br/>
+<em>
+string
+</em>
+</td>
+<td>
+<em>(Optional)</em>
+<p>The data pool name for the Ceph Filesystem subvolume group layout, if the default CephFS pool is not desired.</p>
+</td>
+</tr>
 </table>
 </td>
 </tr>
@@ -3716,6 +3740,30 @@ reference <a href="https://docs.ceph.com/en/latest/cephfs/fs-volumes/#pinning-su
 only one out of (export, distributed, random) can be set at a time</p>
 </td>
 </tr>
+<tr>
+<td>
+<code>quota</code><br/>
+<em>
+k8s.io/apimachinery/pkg/api/resource.Quantity
+</em>
+</td>
+<td>
+<em>(Optional)</em>
+<p>Quota size of the Ceph Filesystem subvolume group.</p>
+</td>
+</tr>
+<tr>
+<td>
+<code>dataPoolName</code><br/>
+<em>
+string
+</em>
+</td>
+<td>
+<em>(Optional)</em>
+<p>The data pool name for the Ceph Filesystem subvolume group layout, if the default CephFS pool is not desired.</p>
+</td>
+</tr>
 </tbody>
 </table>
 <h3 id="ceph.rook.io/v1.CephFilesystemSubVolumeGroupSpecPinning">CephFilesystemSubVolumeGroupSpecPinning

deploy/charts/rook-ceph/templates/resources.yaml

@@ -7983,6 +7983,9 @@ spec:
             spec:
               description: Spec represents the specification of a Ceph Filesystem SubVolumeGroup
               properties:
+                dataPoolName:
+                  description: The data pool name for the Ceph Filesystem subvolume group layout, if the default CephFS pool is not desired.
+                  type: string
                 filesystemName:
                   description: |-
                     FilesystemName is the name of Ceph Filesystem SubVolumeGroup volume name. Typically it's the name of
@@ -8024,6 +8027,13 @@ spec:
                   x-kubernetes-validations:
                     - message: only one pinning type should be set
                       rule: (has(self.export) && !has(self.distributed) && !has(self.random)) || (!has(self.export) && has(self.distributed) && !has(self.random)) || (!has(self.export) && !has(self.distributed) && has(self.random)) || (!has(self.export) && !has(self.distributed) && !has(self.random))
+                quota:
+                  anyOf:
+                    - type: integer
+                    - type: string
+                  description: Quota size of the Ceph Filesystem subvolume group.
+                  pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
+                  x-kubernetes-int-or-string: true
               required:
                 - filesystemName
               type: object

deploy/examples/crds.yaml

@@ -7977,6 +7977,9 @@ spec:
             spec:
               description: Spec represents the specification of a Ceph Filesystem SubVolumeGroup
               properties:
+                dataPoolName:
+                  description: The data pool name for the Ceph Filesystem subvolume group layout, if the default CephFS pool is not desired.
+                  type: string
                 filesystemName:
                   description: |-
                     FilesystemName is the name of Ceph Filesystem SubVolumeGroup volume name. Typically it's the name of
@@ -8018,6 +8021,13 @@ spec:
                   x-kubernetes-validations:
                     - message: only one pinning type should be set
                       rule: (has(self.export) && !has(self.distributed) && !has(self.random)) || (!has(self.export) && has(self.distributed) && !has(self.random)) || (!has(self.export) && !has(self.distributed) && has(self.random)) || (!has(self.export) && !has(self.distributed) && !has(self.random))
+                quota:
+                  anyOf:
+                    - type: integer
+                    - type: string
+                  description: Quota size of the Ceph Filesystem subvolume group.
+                  pattern: ^(\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))(([KMGTPE]i)|[numkMGTPE]|([eE](\+|-)?(([0-9]+(\.[0-9]*)?)|(\.[0-9]+))))?$
+                  x-kubernetes-int-or-string: true
               required:
                 - filesystemName
               type: object

deploy/examples/subvolumegroup.yaml

@@ -17,3 +17,7 @@ spec:
     distributed: 1            # distributed=<0, 1> (disabled=0)
     # export:                 # export=<0-256> (disabled=-1)
     # random:                 # random=[0.0, 1.0](disabled=0.0)
+  # Quota size of the subvolume group.
+  #quota: 10G
+  # data pool name for the subvolume group layout instead of the default data pool.
+  #dataPoolName: myfs-replicated

pkg/apis/ceph.rook.io/v1/types.go

@@ -3022,6 +3022,12 @@ type CephFilesystemSubVolumeGroupSpec struct {
 	// only one out of (export, distributed, random) can be set at a time
 	// +optional
 	Pinning CephFilesystemSubVolumeGroupSpecPinning `json:"pinning,omitempty"`
+	// Quota size of the Ceph Filesystem subvolume group.
+	// +optional
+	Quota *resource.Quantity `json:"quota,omitempty"`
+	// The data pool name for the Ceph Filesystem subvolume group layout, if the default CephFS pool is not desired.
+	// +optional
+	DataPoolName string `json:"dataPoolName"`
 }
 
 // CephFilesystemSubVolumeGroupSpecPinning represents the pinning configuration of SubVolumeGroup

pkg/daemon/ceph/client/subvolumegroup.go

@@ -17,9 +17,11 @@ limitations under the License.
 package client
 
 import (
+	"encoding/json"
 	"fmt"
 	"strconv"
 	"strings"
+	"syscall"
 
 	"github.com/pkg/errors"
 	cephv1 "github.com/rook/rook/pkg/apis/ceph.rook.io/v1"
@@ -30,21 +32,89 @@ import (
 
 // CreateCephFSSubVolumeGroup create a CephFS subvolume group.
 // volName is the name of the Ceph FS volume, the same as the CephFilesystem CR name.
-func CreateCephFSSubVolumeGroup(context *clusterd.Context, clusterInfo *ClusterInfo, volName, groupName string) error {
+func CreateCephFSSubVolumeGroup(context *clusterd.Context, clusterInfo *ClusterInfo, volName, groupName string, svgSpec *cephv1.CephFilesystemSubVolumeGroupSpec) error {
 	logger.Infof("creating cephfs %q subvolume group %q", volName, groupName)
-	//  [--pool_layout <data_pool_name>] [--uid <uid>] [--gid <gid>] [--mode <octal_mode>]
+	// [<size:int>] [--pool_layout <data_pool_name>] [--uid <uid>] [--gid <gid>] [--mode <octal_mode>]
 	args := []string{"fs", "subvolumegroup", "create", volName, groupName}
+	if svgSpec != nil {
+		if svgSpec.Quota != nil {
+			// convert the size to bytes as ceph expect the size in bytes
+			args = append(args, fmt.Sprintf("--size=%d", svgSpec.Quota.Value()))
+		}
+		if svgSpec.DataPoolName != "" {
+			args = append(args, fmt.Sprintf("--pool_layout=%s", svgSpec.DataPoolName))
+		}
+	}
+
+	svgInfo, err := getCephFSSubVolumeGroupInfo(context, clusterInfo, volName, groupName)
+	if err != nil {
+		// return error other than not found.
+		if code, ok := exec.ExitStatus(err); ok && code != int(syscall.ENOENT) {
+			return errors.Wrapf(err, "failed to create subvolume group %q in filesystem %q", groupName, volName)
+		}
+	}
+
+	// if the subvolumegroup exists, resize the subvolumegroup
+	if err == nil && svgSpec != nil && svgSpec.Quota != nil && svgSpec.Quota.CmpInt64(svgInfo.BytesQuota) != 0 {
+		err = resizeCephFSSubVolumeGroup(context, clusterInfo, volName, groupName, svgSpec)
+		if err != nil {
+			return errors.Wrapf(err, "failed to create subvolume group %q in filesystem %q", groupName, volName)
+		}
+	}
+
 	cmd := NewCephCommand(context, clusterInfo, args)
 	cmd.JsonOutput = false
 	output, err := cmd.Run()
 	if err != nil {
-		return errors.Wrapf(err, "failed to create subvolume group %q. %s", volName, output)
+		return errors.Wrapf(err, "failed to create subvolume group %q in filesystem %q. %s", groupName, volName, output)
 	}
 
-	logger.Infof("successfully created cephfs %q subvolume group %q", volName, groupName)
+	logger.Infof("successfully created subvolume group %q in filesystem %q", groupName, volName)
 	return nil
 }
 
+// resizeCephFSSubVolumeGroup resize a CephFS subvolume group.
+// volName is the name of the Ceph FS volume, the same as the CephFilesystem CR name.
+func resizeCephFSSubVolumeGroup(context *clusterd.Context, clusterInfo *ClusterInfo, volName, groupName string, svgSpec *cephv1.CephFilesystemSubVolumeGroupSpec) error {
+	logger.Infof("resizing cephfs %q subvolume group %q", volName, groupName)
+	// <vol_name> <group_name> <new_size> [--no-shrink]
+	args := []string{"fs", "subvolumegroup", "resize", volName, groupName, "--no-shrink", fmt.Sprintf("%d", svgSpec.Quota.Value())}
+	cmd := NewCephCommand(context, clusterInfo, args)
+	cmd.JsonOutput = false
+	output, err := cmd.Run()
+	if err != nil {
+		return errors.Wrapf(err, "failed to resize subvolume group %q in filesystem %q. %s", groupName, volName, output)
+	}
+
+	logger.Infof("successfully resized subvolume group %q in filesystem %q to %s", groupName, volName, svgSpec.Quota)
+	return nil
+}
+
+type subvolumeGroupInfo struct {
+	BytesQuota int64  `json:"bytes_quota"`
+	BytesUsed  int64  `json:"bytes_used"`
+	DataPool   string `json:"data_pool"`
+}
+
+// getCephFSSubVolumeGroupInfo get subvolumegroup info of the group name.
+// volName is the name of the Ceph FS volume, the same as the CephFilesystem CR name.
+func getCephFSSubVolumeGroupInfo(context *clusterd.Context, clusterInfo *ClusterInfo, volName, groupName string) (*subvolumeGroupInfo, error) {
+	args := []string{"fs", "subvolumegroup", "info", volName, groupName}
+	cmd := NewCephCommand(context, clusterInfo, args)
+	cmd.JsonOutput = true
+	output, err := cmd.Run()
+	if err != nil {
+		return nil, errors.Wrapf(err, "failed to get subvolume group %q in filesystem %q. %s", groupName, volName, output)
+	}
+
+	svgInfo := subvolumeGroupInfo{}
+	err = json.Unmarshal(output, &svgInfo)
+	if err != nil {
+		return nil, errors.Wrapf(err, "failed to unmarshal into subvolumeGroupInfo")
+	}
+	return &svgInfo, nil
+}
+
 // DeleteCephFSSubVolumeGroup delete a CephFS subvolume group.
 func DeleteCephFSSubVolumeGroup(context *clusterd.Context, clusterInfo *ClusterInfo, volName, groupName string) error {
 	logger.Infof("deleting cephfs %q subvolume group %q", volName, groupName)

pkg/operator/ceph/file/filesystem.go

@@ -74,7 +74,7 @@ func createFilesystem(
 		}
 	}
 
-	err := cephclient.CreateCephFSSubVolumeGroup(context, clusterInfo, fs.Name, defaultCSISubvolumeGroup)
+	err := cephclient.CreateCephFSSubVolumeGroup(context, clusterInfo, fs.Name, defaultCSISubvolumeGroup, nil)
 	if err != nil {
 		return errors.Wrapf(err, "failed to create subvolume group %q", defaultCSISubvolumeGroup)
 	}

pkg/operator/ceph/file/filesystem_test.go

@@ -233,6 +233,8 @@ func fsExecutor(t *testing.T, fsName, configDir string, multiFS bool, createData
 				return `{"standbys":[], "filesystems":[]}`, nil
 			} else if reflect.DeepEqual(args[0:5], []string{"fs", "subvolumegroup", "create", fsName, defaultCSISubvolumeGroup}) {
 				return "", nil
+			} else if reflect.DeepEqual(args[0:5], []string{"fs", "subvolumegroup", "info", fsName, defaultCSISubvolumeGroup}) {
+				return "", nil
 			} else if contains(args, "osd") && contains(args, "lspools") {
 				return "[]", nil
 			} else if contains(args, "mds") && contains(args, "fail") {

pkg/operator/ceph/file/subvolumegroup/controller.go

@@ -318,7 +318,7 @@ func (r *ReconcileCephFilesystemSubVolumeGroup) updateClusterConfig(cephFilesyst
 func (r *ReconcileCephFilesystemSubVolumeGroup) createOrUpdateSubVolumeGroup(cephFilesystemSubVolumeGroup *cephv1.CephFilesystemSubVolumeGroup) error {
 	logger.Infof("creating ceph filesystem subvolume group %s in namespace %s", cephFilesystemSubVolumeGroup.Name, cephFilesystemSubVolumeGroup.Namespace)
 
-	err := cephclient.CreateCephFSSubVolumeGroup(r.context, r.clusterInfo, cephFilesystemSubVolumeGroup.Spec.FilesystemName, getSubvolumeGroupName(cephFilesystemSubVolumeGroup))
+	err := cephclient.CreateCephFSSubVolumeGroup(r.context, r.clusterInfo, cephFilesystemSubVolumeGroup.Spec.FilesystemName, getSubvolumeGroupName(cephFilesystemSubVolumeGroup), &cephFilesystemSubVolumeGroup.Spec)
 	if err != nil {
 		return errors.Wrapf(err, "failed to create ceph filesystem subvolume group %q", cephFilesystemSubVolumeGroup.Name)
 	}

tests/framework/installer/ceph_manifests.go

@@ -652,7 +652,9 @@ metadata:
   name: ` + groupName + `
   namespace: ` + m.settings.Namespace + `
 spec:
-  filesystemName: ` + fsName
+  filesystemName: ` + fsName + `
+  quota: 10G
+  dataPoolName: ` + fsName + "-data0"
 }
 
 func (m *CephManifestsMaster) GetCOSIDriver() string {