Add job-specific timeouts to GHA test jobs (#20730)

## Summary
Adds timeout commands to all CI test scripts to prevent indefinite hangs, following the pattern established in rapidsai/cuml#7533.

## Changes
Adds `timeout` commands to test execution in 13 CI scripts:
- C++ tests: 30m (run_cudf_ctests, run_cudf_examples, run_cudf_kafka_ctests, run_cudf_benchmark_smoketests)
- C++ memcheck: 2h (observed max: 70 min)
- Java tests: 30m (observed max: 9 min)
- Python cudf tests: 40m each (observed max: 14 min)
- Python other tests: 30m each (observed max: 13 min)
- Wheel cudf tests: 30m each (observed max: 14 min)
- Wheel cudf-polars tests: 1h (observed max: 30 min)
- Wheel dask-cudf tests: 15m (observed max: 6 min)
- cudf-polars polars tests: 30m (observed max: 13 min)
- cudf-polars with rapidsmpf: 15m (observed max: 6 min)
- narwhals tests: 15m each (observed max: 4 min)
- notebooks tests: 10m per notebook (observed max: 3 min)
- third-party integration tests: 45m (observed max: 23 min)

## Timeout Selection
Timeout values are set at 2-3x observed runtimes from recent successful test runs (analyzed from run 19673854267 and others), providing sufficient safety margin while preventing resource waste from hung tests.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - GALI PREM SAGAR (https://github.com/galipremsagar)
  - Gil Forsyth (https://github.com/gforsyth)

URL: #20730

Author: bdice

ci/cudf_pandas_scripts/third-party-integration/test.sh

@@ -85,7 +85,10 @@ main() {
         trap "EXITCODE=1" ERR
         set +e
 
-        TEST_DIR=${TEST_DIR} NUM_PROCESSES=${NUM_PROCESSES} ci/cudf_pandas_scripts/third-party-integration/run-library-tests.sh "${lib}"
+        TEST_DIR=${TEST_DIR} \
+        NUM_PROCESSES=${NUM_PROCESSES} \
+            timeout 45m \
+            ci/cudf_pandas_scripts/third-party-integration/run-library-tests.sh "${lib}"
 
         set -e
         rapids-logger "Test script exiting with value: ${EXITCODE}"

ci/test_cpp.sh

@@ -17,26 +17,26 @@ set +e
 export GTEST_OUTPUT=xml:${RAPIDS_TESTS_DIR}/
 
 rapids-logger "Run libcudf gtests"
-./ci/run_cudf_ctests.sh -j20
+timeout 30m ./ci/run_cudf_ctests.sh -j20
 SUITEERROR=$?
 
 if (( SUITEERROR == 0 )); then
     rapids-logger "Run libcudf examples"
-    ./ci/run_cudf_examples.sh
+    timeout 30m ./ci/run_cudf_examples.sh
     SUITEERROR=$?
 fi
 
 if (( SUITEERROR == 0 )); then
     rapids-logger "Run libcudf_kafka gtests"
-    ./ci/run_cudf_kafka_ctests.sh -j20
+    timeout 30m ./ci/run_cudf_kafka_ctests.sh -j20
     SUITEERROR=$?
 fi
 
 # Ensure that benchmarks are runnable
 rapids-logger "Run tests of libcudf benchmarks"
 
 if (( SUITEERROR == 0 )); then
-    ./ci/run_cudf_benchmark_smoketests.sh
+    timeout 30m ./ci/run_cudf_benchmark_smoketests.sh
     SUITEERROR=$?
 fi
 

ci/test_cpp_memcheck.sh

@@ -11,7 +11,7 @@ source ./ci/test_cpp_common.sh
 
 rapids-logger "Memcheck gtests with rmm_mode=cuda"
 
-./ci/run_cudf_memcheck_ctests.sh && EXITCODE=$? || EXITCODE=$?;
+timeout 2h ./ci/run_cudf_memcheck_ctests.sh && EXITCODE=$? || EXITCODE=$?;
 
 rapids-logger "Test script exiting with value: $EXITCODE"
 # shellcheck disable=SC2086

ci/test_cudf_polars_polars_tests.sh

@@ -49,7 +49,7 @@ trap set_exitcode ERR
 set +e
 
 rapids-logger "Run polars tests"
-./ci/run_cudf_polars_polars_tests.sh
+timeout 30m ./ci/run_cudf_polars_polars_tests.sh
 
 trap ERR
 set -e

ci/test_cudf_polars_with_rapidsmpf.sh

@@ -44,7 +44,7 @@ set +e
 rapids-logger "Running cudf_polars tests with rapidsmpf"
 
 # Run cudf_polars tests with rapidsmpf using dedicated test runner
-./ci/run_cudf_polars_with_rapidsmpf_pytests.sh \
+timeout 15m ./ci/run_cudf_polars_with_rapidsmpf_pytests.sh \
     --no-cov \
     --numprocesses=8 \
     --dist=worksteal \

ci/test_java.sh

@@ -42,7 +42,7 @@ export LIBCUDF_LARGE_STRINGS_ENABLED=0
 
 rapids-logger "Run Java tests"
 pushd java
-mvn test -B -DCUDF_JNI_ENABLE_PROFILING=OFF
+timeout 30m mvn test -B -DCUDF_JNI_ENABLE_PROFILING=OFF
 popd
 
 rapids-logger "Test script exiting with value: $EXITCODE"

ci/test_narwhals.sh

@@ -38,7 +38,9 @@ test_fill_null_series_limit_as_none[cudf] \
 "
 
 rapids-logger "Run narwhals tests for cuDF"
-PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python -m pytest \
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 \
+    timeout 15m \
+    python -m pytest \
     --cache-clear \
     -p xdist \
     -p env \
@@ -65,6 +67,7 @@ CUDF_POLARS__EXECUTOR__TARGET_PARTITION_SIZE=805306368 \
 CUDF_POLARS__EXECUTOR__FALLBACK_MODE=silent \
 PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 \
 NARWHALS_POLARS_GPU=1 \
+    timeout 15m \
     python -m pytest \
     --cache-clear \
     --junitxml="${RAPIDS_TESTS_DIR}/junit-cudf-polars-narwhals.xml" \
@@ -124,7 +127,10 @@ test_explode_multiple_cols or \
 (test_get_dtype_backend and pyarrow and (pandas or modin)) \
 "
 
-PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 NARWHALS_DEFAULT_CONSTRUCTORS=pandas python -m pytest \
+PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 \
+NARWHALS_DEFAULT_CONSTRUCTORS=pandas \
+    timeout 15m \
+    python -m pytest \
     -p cudf.pandas \
     --cache-clear \
     --junitxml="${RAPIDS_TESTS_DIR}/junit-cudf-pandas-narwhals.xml" \

ci/test_notebooks.sh

@@ -58,7 +58,7 @@ for nb in $(find . -name "*.ipynb"); do
         echo "--------------------------------------------------------------------------------"
     else
         nvidia-smi
-        ${NBTEST} "${nbBasename}"
+        timeout 10m ${NBTEST} "${nbBasename}"
     fi
 done
 

ci/test_python_cudf.sh

@@ -18,7 +18,7 @@ trap "EXITCODE=1" ERR
 set +e
 
 rapids-logger "pytest pylibcudf"
-./ci/run_pylibcudf_pytests.sh \
+timeout 40m ./ci/run_pylibcudf_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-pylibcudf.xml" \
   --numprocesses=8 \
   --dist=worksteal \
@@ -28,7 +28,7 @@ rapids-logger "pytest pylibcudf"
   --cov-report=term
 
 rapids-logger "pytest cudf"
-./ci/run_cudf_pytests.sh \
+timeout 40m ./ci/run_cudf_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-cudf.xml" \
   --numprocesses=8 \
   --dist=worksteal \
@@ -43,7 +43,7 @@ rapids-logger "pytest cudf"
 # They do not generate meaningful performance measurements.
 
 rapids-logger "pytest for cudf benchmarks"
-./ci/run_cudf_pytest_benchmarks.sh \
+timeout 40m ./ci/run_cudf_pytest_benchmarks.sh \
   --benchmark-disable \
   --numprocesses=8 \
   --dist=worksteal \
@@ -53,7 +53,7 @@ rapids-logger "pytest for cudf benchmarks"
   --cov-report=term
 
 rapids-logger "pytest for cudf benchmarks using pandas"
-./ci/run_cudf_pandas_pytest_benchmarks.sh \
+timeout 40m ./ci/run_cudf_pandas_pytest_benchmarks.sh \
   --benchmark-disable \
   --numprocesses=8 \
   --dist=worksteal \

ci/test_python_other.sh

@@ -18,7 +18,7 @@ trap "EXITCODE=1" ERR
 set +e
 
 rapids-logger "pytest dask_cudf"
-./ci/run_dask_cudf_pytests.sh \
+timeout 30m ./ci/run_dask_cudf_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-dask-cudf.xml" \
   --numprocesses=8 \
   --dist=worksteal \
@@ -28,11 +28,11 @@ rapids-logger "pytest dask_cudf"
   --cov-report=term
 
 rapids-logger "pytest cudf_kafka"
-./ci/run_cudf_kafka_pytests.sh \
+timeout 30m ./ci/run_cudf_kafka_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-cudf-kafka.xml"
 
 rapids-logger "pytest custreamz"
-./ci/run_custreamz_pytests.sh \
+timeout 30m ./ci/run_custreamz_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-custreamz.xml" \
   --numprocesses=8 \
   --dist=worksteal \
@@ -42,7 +42,7 @@ rapids-logger "pytest custreamz"
   --cov-report=term
 
 rapids-logger "pytest cudf-polars"
-./ci/run_cudf_polars_pytests.sh \
+timeout 30m ./ci/run_cudf_polars_pytests.sh \
   --junitxml="${RAPIDS_TESTS_DIR}/junit-cudf-polars.xml" \
   --numprocesses=8 \
   --dist=worksteal \