When getNamespacesToList returns "" (cross-namespace listing), resources
from all namespaces are listed. The nsTracker.track call for each
resource's namespace would inadvertently track excluded namespaces,
causing their namespace objects to pass through filterNamespaces.
Add a ShouldInclude check before tracking a namespace from non-namespace
resources to ensure excluded namespaces are not tracked.
Add test case for includedNamespaces ["*"] with excludedNamespaces to
verify both the namespace object exclusion and resource exclusion.
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
Commit 8ac8f49b5 ("Remove wildcard check from getNamespacesToList")
removed the optimization that prevented "*" from being expanded to
individual namespace names. This caused getNamespacesToList to return
all namespace names instead of "" (cross-namespace listing), resulting
in N separate API list calls per resource type instead of 1.
On clusters with many namespaces (e.g. 178 on an ACM cluster), this
means ~35,000 API calls instead of ~200, causing backups to take
18-20 minutes for just 8 items.
Restore the "*" special case in ShouldExpandWildcards so that plain
"*" is not expanded, and restore the ShouldInclude("*") check in
getNamespacesToList so that cross-namespace listing is used.
The restore fix from 8ac8f49b5 (fromBackup flag) is preserved since
restores already return false before reaching the "*" check.
Namespace exclusion continues to work correctly: the nsTracker filters
excluded namespace objects via ShouldInclude, and backupItem filters
namespace-scoped resources at line 124 of item_backupper.go.
Fixes#9869
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
Guard the cleanupStubVGSC() call in restore finalization with a check
for VolumeGroupSnapshotHandle in volumeInfo. This avoids a spurious
warning on clusters where the v1beta2 VolumeGroupSnapshotContent CRD
is not installed, since the List call would fail even though no stubs
exist to clean up.
Fixes#9882
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
Per review feedback on #9791, the previous revision still let a
DataUpload with an empty velero.io/backup-name label fall through to
genConfigmap, creating a ConfigMap that deleteMovedSnapshots can never
match back to a snapshot. The CM is useless and only adds etcd churn.
Treat the missing-label case the same way as the foreign-owner case:
warn and skip the ConfigMap creation. Use a distinct warn message so
operators can tell the two misconfiguration classes apart in logs
(missing-label vs. owner mismatch from a captured velero namespace).
Test for the missing-label case is updated to assert no ConfigMap is
created and a warn is emitted. The warn assertion is generalized to
match the per-case message substring instead of a fixed string.
Signed-off-by: Christian Schlichtherle <christian@schlichtherle.de>
Velero does not support self-protection: the velero namespace must
never be captured in a backup tarball. When it is, the tarball can
contain DataUpload CRs belonging to other backups, and the previous
revision of this change silently swallowed that case in the
DataUploadDeleteAction.
Per maintainer feedback, the action should make the misconfiguration
detectable rather than silent. Emit a warn-level log naming the
DataUpload, its owning backup-name label, and the executing backup,
and call out that the velero namespace should be excluded from
schedules. Continue to skip the snapshot-info ConfigMap creation so
that a mislabeled CM does not mask the real owning backup's snapshot
on deletion.
The test for the foreign-backup case now also asserts the warn is
emitted via a logrus test hook.
Signed-off-by: Christian Schlichtherle <cs@bsure-analytics.de>
When a backup tarball incidentally contains DataUpload CRs that belong to
a different backup (common when a schedule includes the velero namespace
where DataUploads live), DataUploadDeleteAction.Execute used to create a
"<du-name>-info" ConfigMap labeled with the *executing* backup's name
instead of the DataUpload's true owning backup. The ConfigMap is
created with Create-only semantics, so the wrong label is never
corrected.
deleteMovedSnapshots in the backup-deletion controller looks up these
ConfigMaps by velero.io/backup-name to discover which Kopia snapshots
to delete. With the wrong label, the real owning backup's expiry pass
finds no ConfigMaps for its DataUploads and silently leaves their Kopia
snapshots in object storage, leaking data over time.
Fix: in DataUploadDeleteAction.Execute, compare the DataUpload's
velero.io/backup-name label against input.Backup.Name (using
label.GetValidName to handle DNS-1035 truncation for long backup names).
If the label is present and differs, skip the DataUpload entirely; this
prevents the over-eager creation of misnamed ConfigMaps without changing
behavior for DataUploads that legitimately belong to the executing
backup, or for legacy DataUploads with no backup-name label.
Refs: #9472
Signed-off-by: Christian Schlichtherle <cs@bsure-analytics.de>
* Fix UT errors: non-constant format string in call to ...
* Fix linter issues.
* Not touch the k8s client version and the controller-runtime version.
Signed-off-by: Xun Jiang <xun.jiang@broadcom.com>
* Fix TestFindVolumeRestoresForPodLegacy error.
* Stablize the TestExecute case in pvc_action_test.go
* Use toolchain to set the specific version for go.
Signed-off-by: Xun Jiang <xun.jiang@broadcom.com>
* Fix wildcard expansion when includes is empty and excludes has wildcards
When a Backup CR is applied via kubectl with empty includedNamespaces
and a wildcard in excludedNamespaces, ShouldExpandWildcards triggers
expansion. The empty includes expands to nil, but wildcardExpanded is
set to true, causing ShouldInclude to return false for all namespaces.
Populate expanded includes with all active namespaces when the original
includes was empty (meaning "include all") so that the wildcardExpanded
check does not falsely reject everything.
Signed-off-by: Joseph <jvaikath@redhat.com>
* Changelog
Signed-off-by: Joseph <jvaikath@redhat.com>
* Normalize empty includes to * instead of active namespaces list
This ensures consistent behavior between CLI and kubectl-apply paths
for Namespace CR inclusion when excludes contain wildcards.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Joseph <jvaikath@redhat.com>
* Move empty includes normalization to backup controller
Instead of normalizing empty IncludedNamespaces to ["*"] in the
collections layer's ExpandIncludesExcludes, do it earlier in
prepareBackupRequest. This ensures the spec is correct before any
downstream processing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Joseph <jvaikath@redhat.com>
* Update TestProcessBackupCompletions for wildcard normalization
Add IncludedNamespaces: []string{"*"} to all expected BackupSpec
structs, reflecting the new prepareBackupRequest normalization.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Joseph <jvaikath@redhat.com>
* Add checks around empty includenamespaces
Signed-off-by: Joseph <jvaikath@redhat.com>
* gofmt
Signed-off-by: Joseph <jvaikath@redhat.com>
---------
Signed-off-by: Joseph <jvaikath@redhat.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Signed-off-by: Xun Jiang <xun.jiang@broadcom.com>
The `getDataUpload` function in the CSI PVC backup plugin was
previously making a cluster-scoped list query to retrieve DataUpload
CRs. In environments with strict minimum-privilege RBAC, this would
fail with forbidden errors.
This explicitly passes the backup namespace into the `ListOptions`
when calling `crClient.List`, correctly scoping the queries to the
backup's namespace. Unit tests have also been updated to ensure
cross-namespace queries are rejected appropriately.
Signed-off-by: Adam Zhang <adam.zhang@broadcom.com>
Kubernetes 1.34 introduced VolumeGroupSnapshot v1beta2 API and
deprecated v1beta1. Distributions running K8s 1.34+ (e.g. OpenShift
4.21+) have removed v1beta1 VGS CRDs entirely, breaking Velero's
VGS functionality on those clusters.
This change bumps external-snapshotter/client/v8 from v8.2.0 to
v8.4.0 and migrates all VGS API usage from v1beta1 to v1beta2.
The v1beta2 API is structurally compatible - the Spec-level types
(GroupSnapshotHandles, VolumeGroupSnapshotContentSource) are
unchanged. The Status-level change (VolumeSnapshotHandlePairList
replaced by VolumeSnapshotInfoList) does not affect Velero as it
does not directly consume that type.
Fixes#9694
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
Restrict the listing of PodVolumeBackup resources to the specific
restore namespace in both the core restore controller and the pod
volume restore action plugin. This prevents "Forbidden" errors when
Velero is configured with namespace-scoped minimum privileges,
avoiding the need for cluster-scoped list permissions for
PodVolumeBackups.
Fixes: #9681
Signed-off-by: Adam Zhang <adam.zhang@broadcom.com>
* Fix VolumeGroupSnapshot restore failure with Ceph RBD CSI driver (#9516)
* Fix VolumeGroupSnapshot restore on Ceph RBD
This PR fixes two related issues affecting CSI snapshot restore on Ceph RBD:
1. VolumeGroupSnapshot restore fails because Ceph RBD populates
volumeGroupSnapshotHandle on pre-provisioned VSCs, but Velero doesn't
create the required VGSC during restore.
2. CSI snapshot restore fails because VolumeSnapshotClassName is removed
from restored VSCs, preventing the CSI controller from getting
credentials for snapshot verification.
Changes:
- Capture volumeGroupSnapshotHandle during backup as VS annotation
- Create stub VGSC during restore with matching handle in status
- Look up VolumeSnapshotClass by driver and set on restored VSC
Fixes#9512Fixes#9515
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Add changelog for VGS restore fix
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Fix gofmt import order
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Add changelog for VGS restore fix
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Fix import alias corev1 to corev1api per lint config
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Fix: Add snapshot handles to existing stub VGSC and add unit tests
When multiple VolumeSnapshots from the same VolumeGroupSnapshot are
restored, they share the same VolumeGroupSnapshotHandle but have
different individual snapshot handles. This commit:
1. Fixes incomplete logic where existing VGSC wasn't updated with
new snapshot handles (addresses review feedback)
2. Fixes race condition where Create returning AlreadyExists would
skip adding the snapshot handle
3. Adds comprehensive unit tests for ensureStubVGSCExists (5 cases)
and addSnapshotHandleToVGSC (4 cases) functions
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Clean up stub VolumeGroupSnapshotContents during restore finalization
Add cleanup logic for stub VGSCs created during VolumeGroupSnapshot restore.
The stub VGSCs are temporary objects needed to satisfy CSI controller
validation during VSC reconciliation. Once all related VSCs become
ReadyToUse, the stub VGSCs are no longer needed and should be removed.
The cleanup runs in the restore finalizer controller's execute() phase.
Before deleting each VGSC, it polls until all related VolumeSnapshotContents
(correlated by snapshot handle) are ReadyToUse, with a timeout fallback.
Deletion failures and CRD-not-installed scenarios are treated as warnings
rather than errors to avoid failing the restore.
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Fix lint: remove unused nolint directive and simplify cleanupStubVGSC return
The cleanupStubVGSC function only produces warnings (not errors), so
simplify its return signature. Also remove the now-unused nolint:unparam
directive on execute() since warnings are no longer always nil.
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
---------
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
* Rename changelog file to match cherry-pick PR number
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
---------
Signed-off-by: Shubham Pampattiwar <spampatt@redhat.com>
Expand wildcard in namespace filter only for backup scenario.
Restore doesn't need that now, because restore has logic to rely on
IncludeEverything function to check whether cluster-scoped resources
should be restored. Expand wildcard will break the logic.
Signed-off-by: Xun Jiang <xun.jiang@broadcom.com>
o.Namespace is empty when Validate runs (Complete hasn't been called yet),
causing VerifyJSONConfigs to query the default namespace instead of the
intended one. Replace o.Namespace with f.Namespace() in all three ConfigMap
validation calls so the factory's already-resolved namespace is used.
Signed-off-by: Adam Zhang <adam.zhang@broadcom.com>
Shubham Pampattiwar
lyndon-liandGitHub
Xun Jiang/Bruce JiangandGitHub
Christian SchlichtherleandLyndon-Li
Christian SchlichtherleandLyndon-Li
Xun Jiang
Priyansh ChoudharyandGitHub
Adam ZhangandXun Jiang/Bruce Jiang
Scott SeagoandTiger Kaovilai
