Files
cozystack/examples/backups/vmi/cleanup.sh
T
Andrey Kolkov 9cc3a9e459 test(vmi): add VMInstance and standalone VMDisk backup/restore e2e round-trips
The disabled hack/e2e-chainsaw/backup/ suite was a Chainsaw port of the retired
backup.bats, written against the old backup API, so the VMInstance and standalone
VMDisk Velero paths had no e2e coverage. Replace it with two round-trips in the
vminstance suite that drive the validated examples/backups/{vmi,vmdisk} scripts,
so the documented demo and the automated test stay in step.

Both run un-gated in CI:
- vminstance-2-backup-roundtrip boots a tiny cirros VM (u1.nano, 2Gi), backs it
  up via BackupClass + BackupJob, restores in place, and asserts the VM returns
  Running off a disk PVC the restore created: a new uid, Velero's restore-name
  label and the -orig- PVC keepOriginalPVC leaves behind. The label alone is not
  enough, because existingResourcePolicy: update also stamps it on a PVC Velero
  only patched, which moves no data. A guest-side boot proof is unavailable: the
  KubeVirt CR disables serial-console logging and cirros ships no guest-agent.
- vmdisk-2-backup-roundtrip imports a 5Gi disk, backs it up, and restores it in
  place with the same three assertions. The controller prepares the target
  (suspend HelmRelease, delete DataVolume, rename PVC). On a WaitForFirstConsumer
  class the Velero data mover waits for a Pod to select a node for the restored
  PVC, which a detached disk never gets, so the script starts a pause Pod on the
  new PVC, standing in for the VM that would attach it.

The disk storageClass is an override (default replicated; the CI steps pass the
lane's COZY_E2E_STORAGE_CLASS, 'local' on the DRBD-off gate) and the script waits
are bounded so their worst case stays under the 40m op timeout, sized to keep the
whole parallel:1 suite inside the 215m job cap. On failure the catch also dumps
the Velero Restore, DataDownload and DataUpload objects. Each harness stands up a
throwaway in-cluster BSL (the platform cozy-default BSL derives the external S3
endpoint, unroutable in kind); its finally resumes the suspended HelmReleases,
reclaims the keepOriginalPVC orig PVC and its LINSTOR volume, and drains the
Velero backups before deleting the BSL. The example scripts are non-interactive
so the harness can drive them.

Assisted-by: LLM
Signed-off-by: Andrey Kolkov <androndo@gmail.com>
2026-10-13 18:05:45 +04:00

88 lines
3.9 KiB
Bash
Executable File

#!/bin/bash
# Clean up all resources created by the demo. Non-interactive: safe to call from
# the e2e harness's finally block, and idempotent (a no-op when a gated-out run
# created nothing). --ignore-not-found returns success for an absent resource
# but does NOT mask a stuck delete: the delete still blocks and, under set -e,
# a teardown that never settles fails this script instead of leaking state into
# the next run (docs/agents/e2e-testing.md convention 4).
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$SCRIPT_DIR/00-helpers.sh"
TARGET_NAMESPACE="${TARGET_NAMESPACE:-tenant-root-copy}"
print_header "Cleanup Demo Resources"
# The driver suspends the VMInstance/VMDisk HelmReleases while a RestoreJob runs
# and never resumes them (suspendHelmRelease only ever sets suspend=true), so an
# aborted or completed restore leaves them suspended. Resume best-effort BEFORE
# deleting the RestoreJobs — deleting a RestoreJob without resuming strands the
# app suspended (same reasoning as examples/backups/postgres/cleanup.sh).
for hr in vm-instance-test vm-disk-ubuntu-source; do
if kubectl -n "$NAMESPACE" get hr "$hr" >/dev/null 2>&1; then
log_substep "Resuming HelmRelease $hr (no-op if not suspended)..."
kubectl -n "$NAMESPACE" patch hr "$hr" --type=merge \
-p '{"spec":{"suspend":false}}' >/dev/null || true
fi
done
# Namespaced request objects first. Deleting the cozystack BackupJob/RestoreJob
# and Backup CRs cascades to the Velero Backup/Restore the controller created.
log_step "Deleting RestoreJobs..."
kubectl delete restorejob restore-to-copy-test restore-in-place-test \
-n "$NAMESPACE" --ignore-not-found
log_step "Deleting BackupJob and Backup..."
kubectl delete backupjob test-backup -n "$NAMESPACE" --ignore-not-found
kubectl delete backup test-backup -n "$NAMESPACE" --ignore-not-found
separator
log_step "Deleting VMInstance and VMDisk..."
kubectl delete vminstance test -n "$NAMESPACE" --ignore-not-found
kubectl delete vmdisk ubuntu-source -n "$NAMESPACE" --ignore-not-found
# keepOriginalPVC renames the VMInstance's disk PVC to <name>-orig-<hash> with a
# bare ObjectMeta (no labels, no ownerRefs) and flips its PV to Retain. Nothing
# above reclaims it: the app delete does not own it and no label selector matches
# it, so under skipDelete a LINSTOR volume and a Released PV leak on every run.
# List loudly (a failed query must abort, not silently skip the reclaim it exists
# for); grep's no-match stays tolerant.
if ! all_pvcs=$(kubectl -n "$NAMESPACE" get pvc -o name 2>&1); then
log_error "cannot list PVCs to reclaim orphaned disks: $all_pvcs"
exit 1
fi
orig_pvcs=$(printf '%s\n' "$all_pvcs" | grep -E '/vm-disk-ubuntu-source-orig-' || true)
for pvc in $orig_pvcs; do
# Flip the retained PV back to Delete BEFORE removing the PVC, so releasing it
# makes the CSI provisioner call DeleteVolume and reclaim the LINSTOR volume —
# deleting the PV object alone would drop the record and strand the volume. Let
# both the lookup and the patch fail the script: swallowing them would delete
# the PVC anyway and strand the volume with no PVC left to find it by.
pv=$(kubectl -n "$NAMESPACE" get "$pvc" -o jsonpath='{.spec.volumeName}')
log_substep "Reclaiming orphaned ${pvc#*/} and its LINSTOR volume..."
if [[ -n "$pv" ]]; then
kubectl patch pv "$pv" --type=merge \
-p '{"spec":{"persistentVolumeReclaimPolicy":"Delete"}}' >/dev/null
fi
kubectl -n "$NAMESPACE" delete "$pvc" --ignore-not-found
done
separator
# Cluster-scoped BackupClass + strategies.
log_step "Deleting BackupClass and Velero strategies..."
kubectl delete backupclass velero --ignore-not-found
kubectl delete velero.strategy.backups.cozystack.io \
vminstance-strategy vmdisk-strategy --ignore-not-found
separator
log_step "Deleting target namespace..."
kubectl delete namespace "$TARGET_NAMESPACE" --ignore-not-found
separator
log_success "Cleanup complete"