github/krkn
1
0
Fork 0
mirror of https://github.com/krkn-chaos/krkn.git synced 2026-08-25 09:27:36 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: serialize node network chaos tests to prevent xdist race condition (#1569)

Browse Source 
* fix: serialize node network chaos tests to prevent xdist race condition

Node network chaos tests share mutable tc rules on the same KinD worker
node. Under pytest-xdist parallel execution, concurrent tests interfere
with each other's tc state causing intermittent marker assertion failures.

Pin all TestNodeNetworkChaos tests to a single xdist worker via
xdist_group marker and switch dist mode to loadgroup so the marker is
respected. Other test classes remain freely distributed.

Signed-off-by: ddjain <darjain@redhat.com>
Co-authored-by: Cursor <cursoragent@cursor.com>

* fix: move --dist loadgroup from pytest.ini to invocation sites

Keeping --dist loadgroup in pytest.ini addopts makes pytest fail at
argument parsing when xdist is not installed, bypassing the preflight
fixture's graceful skip message. Move the flag to the three places
that already pass -n (Makefile test/test-fast, CI workflow) and update
the README parallel-run example accordingly.

Signed-off-by: ddjain <darjain@redhat.com>
Co-authored-by: Cursor <cursoragent@cursor.com>

---------

Signed-off-by: ddjain <darjain@redhat.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Darshan Jain
2026-08-14 22:16:18 +05:30
committed by GitHub
co-authored by Cursor
parent 45c11aa5a9
commit 5f3fe38d5a
5 changed files with 6 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/tests_v2.yml
+1 -1
View File
@@ -53,7 +53,7 @@ jobs:
- name: Run tests_v2
run: |
KRKN_TEST_COVERAGE=1 python -m pytest CI/tests_v2/ -v --timeout=300 --reruns=1 --reruns-delay=5 \
--html=CI/tests_v2/report.html -n auto --junitxml=CI/tests_v2/results.xml
--html=CI/tests_v2/report.html -n auto --dist loadgroup --junitxml=CI/tests_v2/results.xml
- name: Upload tests_v2 artifacts
if: always()
CI/tests_v2/Makefile
+2 -2
View File
@@ -75,10 +75,10 @@ preflight:
test: preflight
cd $(REPO_ROOT) && KRKN_TEST_COVERAGE=1 $(PYTHON) -m pytest $(TESTS_DIR)/ -v --timeout=300 --reruns=2 --reruns-delay=10 \
--html=$(TESTS_DIR)/report.html -n auto --junitxml=$(TESTS_DIR)/results.xml
--html=$(TESTS_DIR)/report.html -n auto --dist loadgroup --junitxml=$(TESTS_DIR)/results.xml
test-fast: preflight
cd $(REPO_ROOT) && $(PYTHON) -m pytest $(TESTS_DIR)/ -v -p no:rerunfailures -n auto --timeout=120
cd $(REPO_ROOT) && $(PYTHON) -m pytest $(TESTS_DIR)/ -v -p no:rerunfailures -n auto --dist loadgroup --timeout=120
test-debug: preflight
cd $(REPO_ROOT) && $(PYTHON) -m pytest $(TESTS_DIR)/ -v -s -p no:rerunfailures --timeout=300 \
CI/tests_v2/README.md
+1 -1
View File
@@ -89,7 +89,7 @@ pytest CI/tests_v2/ -v --timeout=300 --reruns=2 --reruns-delay=10 --html=CI/test
### Run in parallel (faster suite)
```bash
pytest CI/tests_v2/ -v -n 4 --timeout=300
pytest CI/tests_v2/ -v -n 4 --dist loadgroup --timeout=300
```
Ephemeral namespaces make tests parallel-safe; use `-n` with the number of workers (e.g. 4).
CI/tests_v2/pytest.ini
+1
View File
@@ -19,4 +19,5 @@ markers =
namespace_deletion: marks a test as a namespace deletion (service_disruption) scenario test
no_workload: skip workload deployment for this test (e.g. negative tests)
order: set test order (pytest-order)
xdist_group: assign tests to a named xdist group so they run on the same worker (no parallel tc conflicts)
junit_family = xunit2
CI/tests_v2/scenarios/node_network_chaos/test_node_network_chaos.py
+1
View File
@@ -40,6 +40,7 @@ TEST_DURATION = 30
@pytest.mark.functional
@pytest.mark.node_network_chaos
@pytest.mark.xdist_group("node_network_chaos")
class TestNodeNetworkChaos(BaseScenarioTest):
"""Node network chaos: packet loss, latency, bandwidth, direction, targeting, safety, cleanup."""