Skip to content

tests: bump last stable, fix two instabilities #925

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jgehrcke merged 3 commits into from
Mar 9, 2026
Merged

tests: bump last stable, fix two instabilities #925

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a64b97e
tests: bump last stable
jgehrcke Mar 7, 2026
3ff4db3
tests: allow for concurrent image build
jgehrcke Mar 7, 2026
5579d16
tests: filter resource slices (fix #902)
jgehrcke Mar 7, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 11 additions & 2 deletions tests/bats/Makefile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ TEST_CHART_VERSION ?= "$(VERSION_GHCR_CHART)"

# The baseline Helm chart to test upgrades from and downgrades to.
TEST_CHART_LASTSTABLE_REPO ?= "oci://ghcr.io/nvidia/k8s-dra-driver-gpu"
TEST_CHART_LASTSTABLE_VERSION ?= "25.8.1-fc69d985-chart"
TEST_CHART_LASTSTABLE_VERSION ?= "25.12.0-0882da87-chart"

# If not "false": the to-be-tested Helm chart is installed from the local
# filesystem (from `deployments/helm/nvidia-dra-driver-gpu`). Make sure
Expand Down Expand Up @@ -136,9 +136,18 @@ endef

default: tests

# Temporary image tag, see
# https://github.com/NVIDIA/k8s-dra-driver-gpu/issues/908. The `$$$$` below
# evaluates to $$ in the recipe-executing shell, which resolves to its process
# ID. With that, concurrent recipe executions use different image tags,
# preventing rare export errors. Note that `docker tag` is atomic (never fails
# with "already exists").
.PHONY: runner-image
runner-image:
docker buildx build . -t $(BATS_IMAGE) -f tests/bats/Dockerfile
TMP_TAG="$(BATS_IMAGE)-tmp-$$$$" && \
docker buildx build . -t $$TMP_TAG -f tests/bats/Dockerfile && \
docker tag $$TMP_TAG $(BATS_IMAGE) && \
docker rmi $$TMP_TAG

# Warning: destructive against currently configured k8s cluster.
#
Expand Down
32 changes: 22 additions & 10 deletions tests/bats/helpers.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -217,21 +217,33 @@ get_device_attrs_from_any_gpu_slice() {
local device_type="$1"
local node_name="$2"
local spath="${BATS_TEST_TMPDIR}/gpu_resource_slice_content"
local slicename
local slicenames

# Get contents of first listed GPU plugin resource slice; dump into file. If a
# node_name was provided, filter for the GPU plugin resource slice on that
# node.
# For debugging, show current set of slices.
kubectl get resourceslice >&2

# Get names of resource slices: either for all gpu plugin resource slices, or
# all for all gpu slices announced by a specific node.
if [ -n "$node_name" ]; then
slicename="$(kubectl get resourceslices.resource.k8s.io | grep 'gpu.nvidia.com' | grep "$node_name" | head -n1 | awk '{print $1}')"
slicenames="$(kubectl get resourceslices.resource.k8s.io | grep 'gpu.nvidia.com' | grep "$node_name" | awk '{print $1}')"
else
slicename="$(kubectl get resourceslices.resource.k8s.io | grep 'gpu.nvidia.com' | head -n1 | awk '{print $1}')"
slicenames="$(kubectl get resourceslices.resource.k8s.io | grep 'gpu.nvidia.com' | awk '{print $1}')"
fi
log "resource slice name: $slicename"

# For debugging, show current set of slices.
kubectl get resourceslice >&2
kubectl get resourceslices.resource.k8s.io -o yaml "${slicename}" > "${spath}"
# Identify a suitable (device-announcing) resource slice. This is relevant
# because on Kubernetes 1.35+ there is one special resource slice per node
# that only contains counters, and no devices. We must make sure that we
# do not use that for device attribute extraction.
local slicename
for slicename in $slicenames; do
kubectl get resourceslices.resource.k8s.io -o yaml "${slicename}" > "${spath}"
if grep -q "devices:" "${spath}"; then
log "identified suitable resource slice: $slicename"
break
fi
log "do not use slice $slicename (does not define devices)"
done

log "wrote resource slice content to: ${spath}"

# Log contents, for https://github.com/NVIDIA/k8s-dra-driver-gpu/issues/902
Expand Down