Improvements for K8s actions (#1477)

Author: Steboss

.github/container/test-axlearn.sh

@@ -243,3 +243,10 @@ echo "Total number of failed tests ${failed}"
 echo "Total number of skipped tests ${skipped}"
 # add those to summary.txt and we're using it for extracting values
 echo "PASSED: ${passed} FAILED: ${failed} SKIPPED: ${skipped}" >> ${LOG_DIRECTORY}/summary.txt
+# send an error if there are any failed tests
+if [ ${failed} -gt 0 ]; then
+    echo "Some tests failed. Check the logs in ${LOG_DIRECTORY} for details."
+    exit 1
+else
+    echo "All tests passed successfully."
+fi

.github/workflows/_ci.yaml

@@ -623,6 +623,7 @@ jobs:
         job-name: ${{ env.JOB_NAME }}
     - name: Download logs from S3
       id: log-s3
+      if: ${{ !cancelled() }}
       run: |
         mkdir -p axlearn-output
         aws s3 cp s3://jax-toolbox-eks-output/axlearn/${{ github.run_id }}/logs/summary.txt axlearn-output/