workflows: fix in-cluster job kubectl wait

Signed-off-by: Nicolas Busseneau <[email protected]>

.github/workflows/gke.yaml

@@ -106,7 +106,17 @@ jobs:
 
       - name: Wait for job
         run: |
-          kubectl -n kube-system wait job/cilium-cli --for=condition=complete --timeout=15m
+          # Background wait for job to complete or timeout
+          kubectl -n kube-system wait job/cilium-cli --for=condition=complete --timeout=15m &
+          complete_pid=$!
+
+          # Background wait for job to fail
+          kubectl -n kube-system wait job/cilium-cli --for=condition=failed && exit 1 &
+          failed_pid=$!
+
+          # Active wait for whichever background wait ends first
+          wait -n $complete_pid $failed_pid
+          exit $?
 
       - name: Post-test information gathering
         if: ${{ !success() }}