chore(bors): merge pull request #777

777: feat(volume/resize): don't allow volume resize for snapshot restore vols r=dsharma-dc a=dsharma-dc

If a volume is expanded after taking a snapshot or a restore created from that snapshot is expanded, then the IOs to the new io-units(block ranges) fails because the lba range falls out of bound for replica's parent device(snapshot). Till the issue is fixed, this check ensures we don't try to expand a volume if there is snapshot present. SPDK issue 3108.

Co-authored-by: Diwakar Sharma <diwakar.sharma@datacore.com>

control-plane/agents/src/bin/core/volume/operations.rs

@@ -191,11 +191,16 @@ impl ResourceResize for OperationGuardArc<VolumeSpec> {
         let spec = self.as_ref().clone();
         let state = registry.volume_state(&request.uuid).await?;
 
-        // Pre-check - Don't allow resize if the volume has snapshots.
+        // Pre-check - Don't allow resize if the volume has snapshots, or the volume is
+        // a snapshot restore volume.
         // NOTE: This is temporary till the fix for handling this in spdk is done.
-        if !registry.specs().snapshots_by_vol(self.uuid()).is_empty() {
+        if !registry.specs().snapshots_by_vol(self.uuid()).is_empty()
+            || spec.content_source.is_some()
+        {
             return Err(SvcError::Internal {
-                details: "Volume can't be resized while it has snapshots".into(),
+                details:
+                    "Volume can't be resized while it has snapshots, or it's a snapshot restore"
+                        .into(),
             });
         }
         // If the volume is published, then we need to resize nexus also along with replicas.