Update e2e workflow and runtime configurations

- Reduced timeout settings for e2e jobs to optimize execution time.
- Added a step to free up disk space in the e2e workflow to prevent potential issues during tests.
- Installed rsyslog in the Kata Dockerfile to enable proper error logging.
- Updated entrypoint script to start rsyslogd if available, enhancing error reporting capabilities.
- Adjusted E2E_TIMEOUT in the verification script for more efficient test execution.

Author: FAUST-BENCHOU

.github/workflows/e2e.yaml

@@ -11,7 +11,7 @@ on:
 jobs:
   e2e:
     runs-on: ubuntu-latest
-    timeout-minutes: 45
+    timeout-minutes: 20
     steps:
       - name: Check out repository
         uses: actions/checkout@v4
@@ -21,6 +21,11 @@ jobs:
         with:
           go-version-file: go.mod
 
+      - name: Free disk space
+        run: |
+          sudo rm -rf /usr/share/dotnet /usr/local/lib/android /opt/ghc /opt/hostedtoolcache/CodeQL
+          df -h /
+
       - name: Install kind
         run: |
           curl -fsSL -o ./kind "https://kind.sigs.k8s.io/dl/v0.27.0/kind-linux-amd64"
@@ -58,7 +63,7 @@ jobs:
 
   e2e-kata:
     runs-on: ubuntu-latest
-    timeout-minutes: 60
+    timeout-minutes: 20
     steps:
       - name: Check out repository
         uses: actions/checkout@v4
@@ -121,8 +126,6 @@ jobs:
         if: steps.kvm.outputs.available == 'true'
         env:
           CLUSTER_NAME: sandboxfleet-kata
-          # Lifecycle test itself is 5m; keep suite budget tight for kata.
-          E2E_TIMEOUT: 12m
         run: ./hack/verify-e2e.sh
 
       - name: Collect e2e diagnostics

build/runtimes/kata/Dockerfile

@@ -10,9 +10,12 @@ FROM ${BASE_IMAGE}
 ARG TARGETARCH
 ARG KATA_VERSION
 
+# rsyslog: kata-shim reports failures via syslog; without /dev/log the real
+# error is masked as "Unix syslog delivery error: unknown".
 RUN apt-get update \
 	&& DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \
 		zstd \
+		rsyslog \
 	&& rm -rf /var/lib/apt/lists/*
 
 RUN case "${TARGETARCH}" in \

build/worker/entrypoint.sh

@@ -3,6 +3,12 @@ set -eu
 
 mkdir -p /var/lib/containerd /run/containerd /opt/cni/bin /etc/cni/net.d
 
+# Optional: present only on the Kata Worker image. Lets kata-shim log real errors.
+if command -v rsyslogd >/dev/null 2>&1; then
+	mkdir -p /var/run
+	rsyslogd
+fi
+
 # Nested sandboxes need forwarding + NAT through the Worker Pod network.
 if [ -w /proc/sys/net/ipv4/ip_forward ]; then
 	echo 1 >/proc/sys/net/ipv4/ip_forward

hack/verify-e2e.sh

@@ -4,7 +4,7 @@ set -euo pipefail
 ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
 CLUSTER_NAME="${CLUSTER_NAME:-sandboxfleet}"
 E2E_KUBE_CONTEXT="${E2E_KUBE_CONTEXT:-kind-${CLUSTER_NAME}}"
-E2E_TIMEOUT="${E2E_TIMEOUT:-20m}"
+E2E_TIMEOUT="${E2E_TIMEOUT:-10m}"
 KUBECONFIG="${KUBECONFIG:-${ROOT}/bin/KUBECONFIG}"
 
 need() {
@@ -41,6 +41,7 @@ export E2E_KUBE_CONTEXT
 export E2E_RUNTIME_HANDLER="${E2E_RUNTIME_HANDLER:-runsc}"
 
 echo "Running e2e against kubeconfig=${KUBECONFIG} context=${E2E_KUBE_CONTEXT} handler=${E2E_RUNTIME_HANDLER}..."
-go test ./test/e2e/ -tags=e2e -count=1 -v -timeout="${E2E_TIMEOUT}"
+# -failfast: first failure (e.g. lifecycle timeout) aborts the rest of the suite.
+go test ./test/e2e/ -tags=e2e -count=1 -v -failfast -timeout="${E2E_TIMEOUT}"
 
 echo "E2E tests passed."