fix(disks): restore disk from snapshot when storage rounds size up

.dmtlint.yaml

@@ -64,3 +64,69 @@ linters-settings:
         - kind: Deployment
           name: dvcr
           container: dvcr-garbage-collection
+  templates:
+    exclude-rules:
+      # These mount-points.yaml directories pre-create mount points for containerd strict mode.
+      # They are consumed by pods that are NOT rendered from this module's Helm templates:
+      # virt-launcher (created per-VM by KubeVirt at runtime), CDI importer/cloner/uploadserver and
+      # dvcr importer/uploader (created on demand by the controllers), and the virt-*/cdi-* components
+      # deployed by virt-operator/cdi-operator. The mount-points linter only inspects Helm pod
+      # controllers, so it cannot see these mountPaths and reports false positives.
+      mount-points:
+        - /auth
+        - /certs
+        - /data
+        - /dev/bus/usb
+        - /dvcr-auth
+        - /dvcr-src-auth
+        - /etc/docker/registry
+        - /etc/libvirt
+        - /etc/podinfo
+        - /etc/ssl/docker
+        - /etc/virt-api/certificates
+        - /etc/virt-controller/certificates
+        - /etc/virt-controller/exportca
+        - /etc/virt-handler/clientcertificates
+        - /etc/virt-handler/servercertificates
+        - /etc/virt-operator/certificates
+        - /etc/virtualization-api-proxy/certificates
+        - /etc/virtualization-api/certificates
+        - /etc/virtualization-audit/certificates
+        - /init/usr/bin
+        - /kubeconfig.local
+        - /opt
+        - /path
+        - /pods
+        - /profile-data
+        - /proxycerts
+        - /run/ca-bundle/cdi-uploadserver-client-signer-bundle
+        - /run/ca-bundle/cdi-uploadserver-signer-bundle
+        - /run/cdi/clone/source
+        - /run/cdi/token/keys
+        - /run/certs/cdi-apiserver-server-cert
+        - /run/certs/cdi-apiserver-signer-bundle
+        - /run/certs/cdi-uploadserver-client-signer
+        - /run/certs/cdi-uploadserver-signer
+        - /run/cilium
+        - /run/kubevirt
+        - /run/kubevirt-libvirt-runtimes
+        - /run/kubevirt-private
+        - /scratch
+        - /shared
+        - /tmp/k8s-webhook-server/serving-certs
+        - /usr/lib/modules
+        - /var/cache/libvirt
+        - /var/lib/kubelet/device-plugins
+        - /var/lib/kubelet/plugins
+        - /var/lib/kubelet/plugins_registry
+        - /var/lib/kubelet/pods
+        - /var/lib/kubevirt
+        - /var/lib/kubevirt-node-labeller
+        - /var/lib/libvirt
+        - /var/lib/libvirt/qemu
+        - /var/lib/libvirt/qemu/nvram
+        - /var/lib/libvirt/swtpm
+        - /var/lib/registry
+        - /var/lib/swtpm-localca
+        - /var/log/libvirt
+        - /var/run/cdi

.github/actions/milestone-changelog/action.yml

@@ -36,6 +36,7 @@ runs:
           vdsnapshot
           vmsnapshot
           vmrestore
+          vmpool
           disks
           vd
           images

.github/scripts/bash/e2e/cleanup-nightly-resources.sh

@@ -23,6 +23,10 @@ source "${SCRIPT_DIR}/common.sh"
 LABEL_SELECTOR="${LABEL_SELECTOR:-test=nightly-e2e}"
 KEEP_HOURS="${KEEP_HOURS:-47}"
 FRIDAY_KEEP_HOURS="${FRIDAY_KEEP_HOURS:-71}"
+# sds-elastic (Ceph) nested clusters are heavy, so they are torn down sooner (~1 day)
+# and are not granted the Friday extension. Matched by storage type in the resource name.
+ELASTIC_KEEP_HOURS="${ELASTIC_KEEP_HOURS:-23}"
+ELASTIC_NAME_PATTERN="${ELASTIC_NAME_PATTERN:-sds-elastic}"
 
 current_date_seconds="$(date -u +%s)"
 
@@ -35,20 +39,28 @@ collect_items_json() {
 
 should_keep() {
   local created_at="$1"
+  local name="$2"
   local resource_created_at_seconds
   local age_seconds
   local weekday_of_day
+  local keep_hours="${KEEP_HOURS}"
+  local friday_keep_hours="${FRIDAY_KEEP_HOURS}"
+
+  if [[ "${name}" == *"${ELASTIC_NAME_PATTERN}"* ]]; then
+    keep_hours="${ELASTIC_KEEP_HOURS}"
+    friday_keep_hours="${ELASTIC_KEEP_HOURS}"
+  fi
 
   resource_created_at_seconds="$(date -d "${created_at}" -u +%s)"
   age_seconds="$(( current_date_seconds - resource_created_at_seconds ))"
   weekday_of_day="$(date -d "${created_at}" -u +%u)"
 
-  if [ "${age_seconds}" -lt "$(( KEEP_HOURS * 3600 ))" ]; then
+  if [ "${age_seconds}" -lt "$(( keep_hours * 3600 ))" ]; then
     echo "keep"
     return 0
   fi
 
-  if [ "${weekday_of_day}" -eq 5 ] && [ "${age_seconds}" -lt "$(( FRIDAY_KEEP_HOURS * 3600 ))" ]; then
+  if [ "${weekday_of_day}" -eq 5 ] && [ "${age_seconds}" -lt "$(( friday_keep_hours * 3600 ))" ]; then
     echo "keep"
     return 0
   fi
@@ -69,7 +81,7 @@ cleanup_kind() {
     created_at="$(echo "${item}" | jq -r '.created_at')"
     [ -z "${name}" ] && continue
 
-    decision="$(should_keep "${created_at}")"
+    decision="$(should_keep "${created_at}" "${name}")"
     if [ "${decision}" = "keep" ]; then
       printf "%-63s %22s\n" "[INFO] Keep ${kind}/${name}:" "created_at ${created_at}"
       continue

.github/scripts/bash/e2e/configure-sds-elastic.sh

@@ -0,0 +1,134 @@
+#!/usr/bin/env bash
+
+# Copyright 2026 Flant JSC
+#
+# 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.
+
+set -Eeuo pipefail
+
+SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
+# shellcheck source=.github/scripts/bash/e2e/wait-sds-elastic.sh
+source "${SCRIPT_DIR}/wait-sds-elastic.sh"
+
+ELASTIC_CLUSTER_NAME=elastic
+ELASTIC_STORAGE_CLASS=nested-ceph-rbd
+# StorageClassMigration needs a second RBD storage class as a migration target.
+ELASTIC_STORAGE_CLASSES=(nested-ceph-rbd nested-ceph-rbd-r3)
+
+# sds-elastic (Ceph via Rook) is Experimental, so enable it and its dependencies
+# (sds-node-configurator, csi-ceph). On the stage profile the modules are absent in
+# the stage registry, so pull them from the deckhouse-prod ModuleSource created by
+# enable-sdn.sh; otherwise use the default deckhouse source.
+apply_module_configs() {
+  local source_field="  source: deckhouse"
+
+  if [ -n "${MODULE_SOURCE_REGISTRY_CFG:-}" ]; then
+    source_field="  source: deckhouse-prod"
+  fi
+
+  kubectl apply -f - <<EOF
+---
+apiVersion: deckhouse.io/v1alpha1
+kind: ModuleConfig
+metadata:
+  name: sds-node-configurator
+spec:
+  enabled: true
+  version: 1
+${source_field}
+---
+apiVersion: deckhouse.io/v1alpha1
+kind: ModuleConfig
+metadata:
+  name: csi-ceph
+spec:
+  enabled: true
+  version: 1
+${source_field}
+  settings:
+    cephfsEnabled: false
+---
+apiVersion: deckhouse.io/v1alpha1
+kind: ModuleConfig
+metadata:
+  name: sds-elastic
+spec:
+  enabled: true
+  version: 1
+${source_field}
+  settings:
+    dataNodes:
+      nodeSelector:
+        node-role.deckhouse.io/worker: ""
+EOF
+}
+
+echo "[INFO] Allow experimental modules (sds-elastic is Experimental)"
+kubectl patch mc deckhouse --type=merge \
+  -p '{"spec":{"settings":{"allowExperimentalModules":true}}}'
+
+d8_queue
+
+if [ -n "${MODULE_SOURCE_REGISTRY_CFG:-}" ]; then
+  echo "[INFO] Apply sds-elastic ModuleConfigs with deckhouse-prod source (stage profile)"
+else
+  echo "[INFO] Apply sds-elastic ModuleConfigs with deckhouse source"
+fi
+apply_module_configs
+
+echo "[INFO] Wait for sds-node-configurator to be ready"
+kubectl wait --for=jsonpath='{.status.phase}'=Ready modules sds-node-configurator --timeout=300s
+
+echo "[INFO] Wait for csi-ceph to be ready"
+kubectl wait --for=jsonpath='{.status.phase}'=Ready modules csi-ceph --timeout=600s
+
+echo "[INFO] Wait for sds-elastic to be ready"
+kubectl wait --for=jsonpath='{.status.phase}'=Ready modules sds-elastic --timeout=600s
+
+echo "[INFO] Wait for raw block devices to be discovered"
+elastic_blockdevices_ready
+
+echo "[INFO] Label consumable block devices as OSDs (app=elastic-osd)"
+for bd in $(kubectl get blockdevices.storage.deckhouse.io -o json | jq -r '.items[] | select(.status.consumable == true) | .metadata.name'); do
+  echo "[INFO] Label blockdevice ${bd} app=elastic-osd"
+  kubectl label blockdevice "${bd}" app=elastic-osd --overwrite
+done
+
+echo "[INFO] Create ElasticCluster (host networking)"
+kubectl apply -f ./elastic-cluster.yaml
+
+echo "[INFO] Wait for ElasticCluster to be ready"
+elastic_cluster_ready "${ELASTIC_CLUSTER_NAME}"
+
+echo "[INFO] Create ElasticStorageClasses (RBD: replica-2 default + replica-3 migration target)"
+kubectl apply -f ./elastic-storage-class.yaml
+
+for sc in "${ELASTIC_STORAGE_CLASSES[@]}"; do
+  echo "[INFO] Wait for StorageClass ${sc} to appear"
+  for i in $(seq 1 30); do
+    if kubectl get storageclass "${sc}" >/dev/null 2>&1; then
+      echo "[SUCCESS] StorageClass ${sc} is present"
+      break
+    fi
+    echo "[INFO] Wait 10s for StorageClass ${sc} (attempt ${i}/30)"
+    sleep 10
+  done
+done
+
+echo "[INFO] Set default cluster storage class to ${ELASTIC_STORAGE_CLASS}"
+kubectl patch mc global --type='json' \
+  -p='[{"op": "replace", "path": "/spec/settings/defaultClusterStorageClass", "value": "'"${ELASTIC_STORAGE_CLASS}"'"}]'
+
+echo "[INFO] Show existing storageclasses and volumesnapshotclasses"
+kubectl get storageclass
+kubectl get volumesnapshotclass || echo "[WARNING] No volumesnapshotclasses found"

.github/scripts/bash/e2e/power-off-nested-vms.sh

@@ -20,9 +20,9 @@ SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
 # shellcheck source=.github/scripts/bash/e2e/common.sh
 source "${SCRIPT_DIR}/common.sh"
 
-# Constants (nested cluster: 1 master + 3 workers x2)
-REQUIRED_MEM_GI=86
-REQUIRED_CPU=26
+# Constants ((nested cluster: 1 master + 3 workers) x3: replicated + nfs + sds-elastic)
+REQUIRED_MEM_GI=129
+REQUIRED_CPU=39
 MIN_MEM_GI_PER_NODE=12
 MIN_CPU_PER_NODE=4
 MIN_NODES_FOR_PLACEMENT=3

.github/scripts/bash/e2e/render-dvp-static-values.sh

@@ -57,6 +57,19 @@ envsubst_variables="$(grep -oE '\$\{[A-Z0-9_]+\}' values.yaml.tmpl | sort -u | t
 envsubst "${envsubst_variables}" \
   < values.yaml.tmpl > values.yaml
 
+# The template defines one additional worker disk. When ADDITIONAL_DISK_COUNT > 1
+# (e.g. sds-elastic runs several Ceph OSDs per node), append the extra disks, each
+# of ADDITIONAL_DISK_SIZE, to every worker node group.
+additional_disk_count="${ADDITIONAL_DISK_COUNT:-1}"
+if (( additional_disk_count > 1 )); then
+  for (( d = 2; d <= additional_disk_count; d++ )); do
+    # ADDITIONAL_DISK_SIZE is exported by the workflow step env and read by yq's env().
+    yq eval --inplace \
+      '(.instances.additionalNodes[] | select(.name == "worker") | .cfg.additionalDisks) += [{"size": env(ADDITIONAL_DISK_SIZE)}]' \
+      values.yaml
+  done
+fi
+
 mkdir -p tmp
 touch tmp/discovered-values.yaml