Merge pull request #688 from red-hat-storage/sync_ds--master

Syncing latest changes from master for rook

Author: travisn

Documentation/CRDs/Cluster/external-cluster/external-cluster.md

@@ -121,25 +121,6 @@ The storageclass is used to create a volume in the pool matching the topology wh
 
 For more details, see the [Topology-Based Provisioning](topology-for-external-mode.md)
 
-### Upgrade Example
-
-1. If consumer cluster doesn't have restricted caps, this will upgrade all the default csi-users (non-restricted):
-
-    ```console
-    python3 create-external-cluster-resources.py --upgrade
-    ```
-
-2. If the consumer cluster has restricted caps:
-Restricted users created using `--restricted-auth-permission` flag need to pass mandatory flags: '`--rbd-data-pool-name`(if it is a rbd user), `--k8s-cluster-name` and `--run-as-user`' flags while upgrading, in case of cephfs users if you have passed `--cephfs-filesystem-name` flag while creating csi-users then while upgrading it will be mandatory too. In this example the user would be `client.csi-rbd-node-rookstorage-replicapool` (following the pattern `csi-user-clusterName-poolName`)
-
-    ```console
-    python3 create-external-cluster-resources.py --upgrade --rbd-data-pool-name replicapool --k8s-cluster-name rookstorage --run-as-user client.csi-rbd-node-rookstorage-replicapool
-    ```
-
-!!! note
-    An existing non-restricted user cannot be converted to a restricted user by upgrading.
-    The upgrade flag should only be used to append new permissions to users. It shouldn't be used for changing a csi user already applied permissions. For example, you shouldn't change the pool(s) a user has access to.
-
 ### Admin privileges
 
 If in case the cluster needs the admin keyring to configure, update the admin key `rook-ceph-mon` secret with client.admin keyring
@@ -305,3 +286,57 @@ you can export the settings from this cluster with the following steps.
 
 !!! important
     For other clusters to connect to storage in this cluster, Rook must be configured with a networking configuration that is accessible from other clusters. Most commonly this is done by enabling host networking in the CephCluster CR so the Ceph daemons will be addressable by their host IPs.
+
+## Upgrades
+
+Upgrading the cluster would be different for restricted caps and non-restricted caps,
+
+1. If consumer cluster doesn't have restricted caps, this will upgrade all the default CSI users (non-restricted)
+
+    ```console
+    python3 create-external-cluster-resources.py --upgrade
+    ```
+
+2. If the consumer cluster has restricted caps
+
+    Restricted users created using `--restricted-auth-permission` flag need to pass mandatory flags: '`--rbd-data-pool-name`(if it is a rbd user), `--k8s-cluster-name` and `--run-as-user`' flags while upgrading, in case of cephfs users if you have passed `--cephfs-filesystem-name` flag while creating CSI users then while upgrading it will be mandatory too. In this example the user would be `client.csi-rbd-node-rookstorage-replicapool` (following the pattern `csi-user-clusterName-poolName`)
+
+    ```console
+    python3 create-external-cluster-resources.py --upgrade --rbd-data-pool-name replicapool --k8s-cluster-name rookstorage --run-as-user client.csi-rbd-node-rookstorage-replicapool
+    ```
+
+    !!! note
+        1) An existing non-restricted user cannot be converted to a restricted user by upgrading.
+        2) The upgrade flag should only be used to append new permissions to users. It shouldn't be used for changing a CSI user already applied permissions. For example, be careful not to change pools(s) that a user has access to.
+
+### Upgrade cluster to utilize new feature
+
+Some Rook upgrades may require re-running the import steps, or may introduce new external cluster features that can be most easily enabled by re-running the import steps.
+
+To re-run the import steps with new options, the python script should be re-run using the same configuration options that were used for past invocations, plus the configurations that are being added or modified.
+
+Starting with Rook v1.15, the script stores the configuration in the external-cluster-user-command configmap for easy future reference.
+
+* arg: Exact arguments that were used for for processing the script. Argument that are decided using the Priority: command-line-args > config.ini file values > default values.
+
+#### Example `external-cluster-user-command` ConfigMap:
+
+1. Get the last-applied config, if its available
+
+    ```console
+    $ kubectl get configmap  -namespace rook-ceph external-cluster-user-command --output jsonpath='{.data.args}'
+    ```
+
+2. Copy the output to config.ini
+
+3. Make any desired modifications and additions to `config.ini``
+
+4. Run the python script again using the [config file](#config-file)
+
+5. [Copy the bash output](#2-copy-the-bash-output)
+
+6. Run the steps under [import-the-source-data](#import-the-source-data)
+
+!!! warning
+    If the last-applied config is unavailable, run the current version of the script again using previously-applied config and CLI flags.
+    Failure to reuse the same configuration options when re-invoking the python script can result in unexpected changes when re-running the import script.

Documentation/Troubleshooting/ceph-common-issues.md

@@ -67,6 +67,9 @@ title: Ceph Common Issues
     - [Symptoms](#symptoms-11)
     - [Investigation](#investigation-7)
     - [Solution](#solution-12)
+- [The cluster is in an unhealthy state or fails to configure when LimitNOFILE=infinity in containerd](#the-cluster-is-in-an-unhealthy-state-or-fails-to-configure-when-limitnofileinfinity-in-containerd)
+    - [Symptoms](#symptoms-12)
+    - [Solution](#solution-13)
 
 
 Many of these problem cases are hard to summarize down to a short phrase that adequately describes the problem. Each problem will start with a bulleted list of symptoms. Keep in mind that all symptoms may not apply depending on the configuration of Rook. If the majority of the symptoms are seen there is a fair chance you are experiencing that problem.
@@ -774,3 +777,36 @@ data: {}
 ```
 
 If the ConfigMap exists, remove any keys that you wish to configure through the environment.
+
+## The cluster is in an unhealthy state or fails to configure when LimitNOFILE=infinity in containerd
+
+### Symptoms
+
+When trying to create a new deployment, Ceph mons keep crashing and the cluster fails to configure or remains in an unhealthy state. The nodes' CPUs are stuck at 100%.
+
+```console
+NAME        DATADIRHOSTPATH   MONCOUNT   AGE    PHASE   MESSAGE                                    HEALTH       EXTERNAL   FSID
+rook-ceph   /var/lib/rook      3         4m6s   Ready   Failed to configure ceph cluster           HEALTH_ERR
+```
+
+### Solution
+
+Before systemd v240, systemd would leave `fs.nr_open` as-is because it had no mechanism to set a safe upper limit for it. The kernel hard-coded value for the default number of max open files is **1048576**. Starting from systemd v240, when `LimitNOFILE=infinity` is specified in the containerd.service configuration, this value will typically be set to **~1073741816** (INT_MAX for x86_64 divided by two).
+
+To fix this, set LimitNOFILE in the systemd service configuration to **1048576**.
+
+Create an override.conf file with the new LimitNOFILE value:
+
+```console
+$ vim /etc/systemd/system/containerd.service.d/override.conf
+[Service]
+LimitNOFILE=1048576
+```
+
+Reload systemd manager configuration, restart containerd and restart all monitors deployments:
+
+```console
+$ systemctl daemon-reload
+$ systemctl restart containerd
+$ kubectl rollout restart deployment rook-ceph-mon-a rook-ceph-mon-b rook-ceph-mon-c -n rook-ceph
+```

build/csv/ceph/ceph.rook.io_cephclusters.yaml

@@ -449,6 +449,7 @@ spec:
                               - ""
                               - crush-compat
                               - upmap
+                              - read
                               - upmap-read
                               type: string
                           type: object

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

@@ -1601,6 +1601,7 @@ spec:
                                   - ""
                                   - crush-compat
                                   - upmap
+                                  - read
                                   - upmap-read
                                 type: string
                             type: object

deploy/examples/crds.yaml

@@ -1599,6 +1599,7 @@ spec:
                                   - ""
                                   - crush-compat
                                   - upmap
+                                  - read
                                   - upmap-read
                                 type: string
                             type: object

go.mod

@@ -16,7 +16,7 @@ replace (
 
 require (
 	github.com/IBM/keyprotect-go-client v0.14.3
-	github.com/aws/aws-sdk-go v1.54.20
+	github.com/aws/aws-sdk-go v1.55.3
 	github.com/banzaicloud/k8s-objectmatcher v1.8.0
 	github.com/ceph/go-ceph v0.28.0
 	github.com/coreos/pkg v0.0.0-20230601102743-20bbbf26f4d8
@@ -30,8 +30,8 @@ require (
 	github.com/kube-object-storage/lib-bucket-provisioner v0.0.0-20221122204822-d1a8c34382f1
 	github.com/libopenstorage/secrets v0.0.0-20240416031220-a17cf7f72c6c
 	github.com/pkg/errors v0.9.1
-	github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.1
-	github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.1
+	github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.2
+	github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.2
 	github.com/rook/rook/pkg/apis v0.0.0-20231204200402-5287527732f7
 	github.com/spf13/cobra v1.8.1
 	github.com/spf13/pflag v1.0.5

go.sum

@@ -144,8 +144,8 @@ github.com/armon/go-socks5 v0.0.0-20160902184237-e75332964ef5/go.mod h1:wHh0iHkY
 github.com/asaskevich/govalidator v0.0.0-20180720115003-f9ffefc3facf/go.mod h1:lB+ZfQJz7igIIfQNfa7Ml4HSf2uFQQRzpGGRXenZAgY=
 github.com/asaskevich/govalidator v0.0.0-20190424111038-f61b66f89f4a/go.mod h1:lB+ZfQJz7igIIfQNfa7Ml4HSf2uFQQRzpGGRXenZAgY=
 github.com/aws/aws-sdk-go v1.44.164/go.mod h1:aVsgQcEevwlmQ7qHE9I3h+dtQgpqhFB+i8Phjh7fkwI=
-github.com/aws/aws-sdk-go v1.54.20 h1:FZ2UcXya7bUkvkpf7TaPmiL7EubK0go1nlXGLRwEsoo=
-github.com/aws/aws-sdk-go v1.54.20/go.mod h1:eRwEWoyTWFMVYVQzKMNHWP5/RV4xIUGMQfXQHfHkpNU=
+github.com/aws/aws-sdk-go v1.55.3 h1:0B5hOX+mIx7I5XPOrjrHlKSDQV/+ypFZpIHOx5LOk3E=
+github.com/aws/aws-sdk-go v1.55.3/go.mod h1:eRwEWoyTWFMVYVQzKMNHWP5/RV4xIUGMQfXQHfHkpNU=
 github.com/banzaicloud/k8s-objectmatcher v1.8.0 h1:Nugn25elKtPMTA2br+JgHNeSQ04sc05MDPmpJnd1N2A=
 github.com/banzaicloud/k8s-objectmatcher v1.8.0/go.mod h1:p2LSNAjlECf07fbhDyebTkPUIYnU05G+WfGgkTmgeMg=
 github.com/benbjohnson/clock v1.1.0/go.mod h1:J11/hYXuz8f4ySSvYwY0FKfm+ezbsZBKZxNJlLklBHA=
@@ -770,11 +770,11 @@ github.com/prashantv/gostub v1.1.0 h1:BTyx3RfQjRHnUWaGF9oQos79AlQ5k8WNktv7VGvVH4
 github.com/prashantv/gostub v1.1.0/go.mod h1:A5zLQHz7ieHGG7is6LLXLz7I8+3LZzsrV0P1IAHhP5U=
 github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.44.1/go.mod h1:3WYi4xqXxGGXWDdQIITnLNmuDzO5n6wYva9spVhR4fg=
 github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.46.0/go.mod h1:3WYi4xqXxGGXWDdQIITnLNmuDzO5n6wYva9spVhR4fg=
-github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.1 h1:+iiljhJV6niK7MuifJs/n3NeLxikd85nrQfn53sLJkU=
-github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.1/go.mod h1:XYrdZw5dW12Cjkt4ndbeNZZTBp4UCHtW0ccR9+sTtPU=
+github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.2 h1:6UsAv+jAevuGO2yZFU/BukV4o9NKnFMOuoouSA4G0ns=
+github.com/prometheus-operator/prometheus-operator/pkg/apis/monitoring v0.75.2/go.mod h1:XYrdZw5dW12Cjkt4ndbeNZZTBp4UCHtW0ccR9+sTtPU=
 github.com/prometheus-operator/prometheus-operator/pkg/client v0.46.0/go.mod h1:k4BrWlVQQsvBiTcDnKEMgyh/euRxyxgrHdur/ZX/sdA=
-github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.1 h1:s7GlsRYGLWP+L1eQKy6RmLatX+k3v9NQwutUix4l5uM=
-github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.1/go.mod h1:qca3qWGdknRpHvPyThepe5a6QYAh38IQ2ml93E6V3NY=
+github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.2 h1:71GOmhZFA2/17maXqCcuJEzpJDyqPty8SpEOGZWyVec=
+github.com/prometheus-operator/prometheus-operator/pkg/client v0.75.2/go.mod h1:Sv6XsfGGkR9gKnhP92F5dNXEpsSePn0W+7JwYP0NVkc=
 github.com/prometheus/client_golang v0.9.0/go.mod h1:7SWBe2y4D6OKWSNQJUaRYU/AaXPKyh/dDVn+NZz0KFw=
 github.com/prometheus/client_golang v0.9.1/go.mod h1:7SWBe2y4D6OKWSNQJUaRYU/AaXPKyh/dDVn+NZz0KFw=
 github.com/prometheus/client_golang v0.9.3/go.mod h1:/TN21ttK/J9q6uSwhBd54HahCDft0ttaMvbicHlPoso=

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

@@ -679,7 +679,7 @@ type Module struct {
 
 type ModuleSettings struct {
 	// BalancerMode sets the `balancer` module with different modes like `upmap`, `crush-compact` etc
-	// +kubebuilder:validation:Enum="";crush-compat;upmap;upmap-read
+	// +kubebuilder:validation:Enum="";crush-compat;upmap;read;upmap-read
 	BalancerMode string `json:"balancerMode,omitempty"`
 }
 

pkg/daemon/ceph/client/mgr.go

@@ -22,12 +22,18 @@ import (
 
 	"github.com/pkg/errors"
 	"github.com/rook/rook/pkg/clusterd"
+	cephver "github.com/rook/rook/pkg/operator/ceph/version"
 )
 
 var (
 	moduleEnableWaitTime = 5 * time.Second
 )
 
+const (
+	readBalancerMode      = "read"
+	upmapReadBalancerMode = "upmap-read"
+)
+
 func CephMgrMap(context *clusterd.Context, clusterInfo *ClusterInfo) (*MgrMap, error) {
 	args := []string{"mgr", "dump"}
 	buf, err := NewCephCommand(context, clusterInfo, args).Run()
@@ -132,12 +138,12 @@ func setBalancerMode(context *clusterd.Context, clusterInfo *ClusterInfo, mode s
 	return nil
 }
 
-// setMinCompatClientLuminous set the minimum compatibility for clients to Luminous
-func setMinCompatClientLuminous(context *clusterd.Context, clusterInfo *ClusterInfo) error {
-	args := []string{"osd", "set-require-min-compat-client", "luminous", "--yes-i-really-mean-it"}
+// setMinCompatClient set the minimum compatibility for clients
+func setMinCompatClient(context *clusterd.Context, clusterInfo *ClusterInfo, version string) error {
+	args := []string{"osd", "set-require-min-compat-client", version, "--yes-i-really-mean-it"}
 	_, err := NewCephCommand(context, clusterInfo, args).Run()
 	if err != nil {
-		return errors.Wrap(err, "failed to set set-require-min-compat-client to luminous")
+		return errors.Wrapf(err, "failed to set set-require-min-compat-client to %q", version)
 	}
 
 	return nil
@@ -165,8 +171,12 @@ func mgrSetBalancerMode(context *clusterd.Context, clusterInfo *ClusterInfo, bal
 
 // ConfigureBalancerModule configures the balancer module
 func ConfigureBalancerModule(context *clusterd.Context, clusterInfo *ClusterInfo, balancerModuleMode string) error {
-	// Set min compat client to luminous before enabling the balancer mode "upmap"
-	err := setMinCompatClientLuminous(context, clusterInfo)
+	minCompatClientVersion, err := desiredMinCompatClientVersion(clusterInfo, balancerModuleMode)
+	if err != nil {
+		return errors.Wrap(err, "failed to get minimum compatibility client version")
+	}
+
+	err = setMinCompatClient(context, clusterInfo, minCompatClientVersion)
 	if err != nil {
 		return errors.Wrap(err, "failed to set minimum compatibility client")
 	}
@@ -179,3 +189,17 @@ func ConfigureBalancerModule(context *clusterd.Context, clusterInfo *ClusterInfo
 
 	return nil
 }
+
+func desiredMinCompatClientVersion(clusterInfo *ClusterInfo, balancerModuleMode string) (string, error) {
+	// Set min compat client to luminous before enabling the balancer mode "upmap"
+	minCompatClientVersion := "luminous"
+	if balancerModuleMode == readBalancerMode || balancerModuleMode == upmapReadBalancerMode {
+		if !clusterInfo.CephVersion.IsAtLeast(cephver.CephVersion{Major: 19}) {
+			return "", errors.New("minimum ceph v19 (Squid) is required for upmap-read or read balancer modes")
+		}
+		// Set min compat client to reef before enabling the balancer mode "upmap-read" or "read"
+		minCompatClientVersion = "reef"
+	}
+
+	return minCompatClientVersion, nil
+}

pkg/daemon/ceph/client/mgr_test.go

@@ -21,6 +21,7 @@ import (
 
 	"github.com/pkg/errors"
 	"github.com/rook/rook/pkg/clusterd"
+	cephver "github.com/rook/rook/pkg/operator/ceph/version"
 	exectest "github.com/rook/rook/pkg/util/exec/test"
 	"github.com/stretchr/testify/assert"
 )
@@ -135,3 +136,37 @@ func TestSetBalancerMode(t *testing.T) {
 	err := setBalancerMode(&clusterd.Context{Executor: executor}, AdminTestClusterInfo("mycluster"), "upmap")
 	assert.NoError(t, err)
 }
+
+func TestGetMinCompatClientVersion(t *testing.T) {
+	clusterInfo := AdminTestClusterInfo("mycluster")
+	t.Run("upmap-read balancer mode with ceph v19", func(t *testing.T) {
+		clusterInfo.CephVersion = cephver.CephVersion{Major: 19}
+		result, err := desiredMinCompatClientVersion(clusterInfo, upmapReadBalancerMode)
+		assert.NoError(t, err)
+		assert.Equal(t, "reef", result)
+	})
+
+	t.Run("read balancer mode with ceph v19", func(t *testing.T) {
+		clusterInfo.CephVersion = cephver.CephVersion{Major: 19}
+		result, err := desiredMinCompatClientVersion(clusterInfo, readBalancerMode)
+		assert.NoError(t, err)
+		assert.Equal(t, "reef", result)
+	})
+	t.Run("upmap-read balancer mode with ceph below v19 should fail", func(t *testing.T) {
+		clusterInfo.CephVersion = cephver.CephVersion{Major: 18}
+		_, err := desiredMinCompatClientVersion(clusterInfo, upmapReadBalancerMode)
+		assert.Error(t, err)
+	})
+	t.Run("read balancer mode with ceph below v19 should fail", func(t *testing.T) {
+		clusterInfo.CephVersion = cephver.CephVersion{Major: 18}
+		_, err := desiredMinCompatClientVersion(clusterInfo, readBalancerMode)
+		assert.Error(t, err)
+	})
+
+	t.Run("upmap balancer set min compat client to luminous", func(t *testing.T) {
+		clusterInfo.CephVersion = cephver.CephVersion{Major: 19}
+		result, err := desiredMinCompatClientVersion(clusterInfo, "upmap")
+		assert.NoError(t, err)
+		assert.Equal(t, "luminous", result)
+	})
+}