Merge pull request #1070 from marquiz/release-0.8

[release-0.8] backport e2e test fixes from master

Makefile

@@ -472,23 +472,7 @@ e2e-tests: build-static
 	fi
 
 packaging-tests: cross-packages
-	$(Q)for dir in test/e2e/packages.test-suite/*; do \
-	    cleanup=1 omit_agent=1 $(E2E_RUN) $$dir; \
-	    echo "--------------------------------------"; \
-	    for r in $(find $$dir -name summary.txt); do \
-	        t="$${r#*.test-suite/}"; t="$${t%/output/summary.txt}"; \
-	        echo "$$t:"; \
-	        cat $$r | sed 's/^/    /g'; \
-	    done; \
-	    echo "======================================"; \
-	done; \
-	for dir in test/e2e/packages.test-suite/*; do \
-	    for r in $(find $$dir -name summary.txt); do \
-	        t="$${r#*.test-suite/}"; t="$${t%/output/summary.txt}"; \
-	        echo "$$t:"; \
-	        cat $$r | sed 's/^/    /g'; \
-	    done; \
-	done;
+	$(Q)cleanup=1 omit_agent=1 $(E2E_RUN) test/e2e/packages.test-suite
 
 #
 # Rules for building distro packages.

demo/lib/distro.bash

@@ -130,7 +130,7 @@ ubuntu-22_04-image-url() {
 }
 
 debian-10-image-url() {
-    echo "https://cloud.debian.org/images/cloud/buster/20200803-347/debian-10-generic-amd64-20200803-347.qcow2"
+    echo "https://cloud.debian.org/images/cloud/buster/latest/debian-10-generic-amd64.qcow2"
 }
 
 debian-11-image-url() {
@@ -381,21 +381,17 @@ centos-8-install-containerd-pre() {
     distro-install-repo https://download.docker.com/linux/centos/docker-ce.repo
 }
 
-centos-7-k8s-cni() {
-    echo "weavenet"
-}
-
 centos-install-golang() {
     distro-install-pkg wget tar gzip git-core
     from-tarball-install-golang
 }
 
 fedora-image-url() {
-    fedora-36-image-url
+    fedora-38-image-url
 }
 
-fedora-36-image-url() {
-    echo "https://mirrors.xtom.de/fedora/releases/36/Cloud/x86_64/images/Fedora-Cloud-Base-36-1.5.x86_64.qcow2"
+fedora-38-image-url() {
+    echo "https://mirrors.xtom.de/fedora/releases/38/Cloud/x86_64/images/Fedora-Cloud-Base-38-1.6.x86_64.qcow2"
 }
 
 fedora-35-image-url() {
@@ -686,7 +682,11 @@ opensuse-image-url() {
 }
 
 opensuse-15_4-image-url() {
-    echo "https://download.opensuse.org/pub/opensuse/distribution/leap/15.4/appliances/openSUSE-Leap-15.4-JeOS.x86_64-15.4-OpenStack-Cloud-Current.qcow2"
+    echo "https://download.opensuse.org/pub/opensuse/distribution/leap/15.4/appliances/openSUSE-Leap-15.4-Minimal-VM.x86_64-OpenStack-Cloud.qcow2"
+}
+
+opensuse-15_5-image-url() {
+    echo "https://download.opensuse.org/pub/opensuse/distribution/leap/15.5/appliances/openSUSE-Leap-15.5-Minimal-VM.x86_64-Cloud.qcow2"
 }
 
 opensuse-tumbleweed-image-url() {
@@ -781,7 +781,9 @@ opensuse-wait-for-zypper() {
 opensuse-require-repo-virtualization-containers() {
     vm-command "zypper ls"
     if ! grep -q Virtualization_containers <<< "$COMMAND_OUTPUT"; then
-        opensuse-install-repo https://download.opensuse.org/repositories/Virtualization:containers/15.4/Virtualization:containers.repo
+        vm-command 'source /etc/os-release; echo $VERSION'
+        local opensuse_version=$COMMAND_OUTPUT
+        opensuse-install-repo https://download.opensuse.org/repositories/Virtualization:containers/${opensuse_version}/Virtualization:containers.repo
         opensuse-refresh-pkg-db
     fi
 }
@@ -1053,7 +1055,7 @@ default-install-utils() {
 }
 
 default-k8s-cni() {
-    echo ${k8scni:-cilium}
+    echo ${k8scni:-bridge}
 }
 
 default-k8s-cni-subnet() {

demo/lib/topology.py

@@ -54,7 +54,7 @@
 
 _bash_topology_dump = """for cpu in /sys/devices/system/cpu/cpu[0-9]*; do cpu_id=${cpu#/sys/devices/system/cpu/cpu}; echo "cpu p:$(< ${cpu}/topology/physical_package_id) d:$(< ${cpu}/topology/die_id) n:$(basename  ${cpu}/node* | sed 's:node::g') c:$(< ${cpu}/topology/core_id) t:$(< ${cpu}/topology/thread_siblings) cpu:${cpu_id}" ; done;  for node in /sys/devices/system/node/node[0-9]*; do node_id=${node#/sys/devices/system/node/node}; echo "dist n:$node_id d:$(< $node/distance)"; echo "mem n:$node_id s:$(awk '/MemTotal/{print $4/1024}' < $node/meminfo)"; done"""
 
-_bash_res_allowed = r"""for process in '%s'; do for pid in $(pgrep -f "$process"); do name=$(cat /proc/$pid/cmdline | tr '\0 ' '\n' | grep -E "^$process" | head -n 1); [ -n "$name" ] && [ "$pid" != "$$" ] && [ "$pid" != "$PPID" ] && echo "${name}/${pid} $(awk '/Cpus_allowed:/{c=$2}/Mems_allowed:/{m=$2}END{print "c:"c" m:"m}' < /proc/$pid/status)"; done; done"""
+_bash_res_allowed = r"""for process in '%s'; do for pid in $(pgrep -f "$process"); do proc_pid_cmdline=$(< /proc/$pid/cmdline) || continue; proc_pid_status=$(< /proc/$pid/status) || continue; name=$(echo "$proc_pid_cmdline" | tr '\0 ' '\n' | grep -E "^$process" | head -n 1); [ -n "$name" ] && [ "$pid" != "$$" ] && [ "$pid" != "$PPID" ] && echo "${name}/${pid} $(awk '/Cpus_allowed:/{c=$2}/Mems_allowed:/{m=$2}END{print "c:"c" m:"m}' <<< "$proc_pid_status")"; done 2>/dev/null; done"""
 
 def error(msg, exit_status=1):
     """Print error message and exit."""

demo/lib/vm.bash

@@ -616,25 +616,10 @@ vm-reboot() { # script API
     # Reboots the virtual machine and waits that the ssh server starts
     # responding again.
     vm-command "reboot"
-    sleep 5
-    host-wait-vm-ssh-server
-}
-
-vm-force-restart() { # script API
-    # Usage: vm-force-restart
-    #
-    # Give the virtual machine a chance to shut itself down, then
-    # forcibly restart it using govm stop/start. Wait for the ssh
-    # server to start responding again after restarting. If VM_NAME
-    # is not set assume the target machine to not be govm-managed
-    # and fall back to vm-reboot instead.
-    if vm-is-govm; then
-        vm-command "shutdown -h now"
-        sleep 10
+    sleep 10
+    if ! host-wait-vm-ssh-server; then
         vm-monitor system_reset
         host-wait-vm-ssh-server
-    else
-      vm-reboot
     fi
 }
 
@@ -1024,6 +1009,7 @@ vm-create-singlenode-cluster() {
     if ! vm-command "kubectl wait --for=condition=Ready node/\$(hostname) --timeout=240s"; then
         command-error "kubectl waiting for node readiness timed out"
     fi
+    vm-run-until --timeout 30 "kubectl get sa default > /dev/null" || error "serviceaccount 'default' not found"
 }
 
 vm-create-cluster() {
@@ -1046,6 +1032,25 @@ vm-destroy-cluster() {
     vm-command "yes | kubeadm reset; rm -f ~$user/.kube/config ~root/.kube/config /etc/kubernetes"
 }
 
+vm-install-cni-bridge() {
+    vm-command "rm -rf /etc/cni/net.d/* && mkdir -p /etc/cni/net.d && cat > /etc/cni/net.d/10-bridge.conf <<EOF
+{
+  \"cniVersion\": \"0.4.0\",
+  \"name\": \"demonet\",
+  \"type\": \"bridge\",
+  \"isGateway\": true,
+  \"ipMasq\": true,
+  \"ipam\": {
+    \"type\": \"host-local\",
+    \"subnet\": \"$CNI_SUBNET\",
+    \"routes\": [
+      { \"dst\": \"0.0.0.0/0\" }
+    ]
+  }
+}
+EOF"
+}
+
 vm-install-cni-cilium() {
     if ! vm-command "curl -L --remote-name-all https://github.com/cilium/cilium-cli/releases/latest/download/cilium-linux-amd64.tar.gz && tar xzvfC cilium-linux-amd64.tar.gz /usr/local/bin && cilium install && rm -f cilium-linux-amd64.tar.gz"; then
         command-error "installing cilium CNI to Kubernetes failed"

test/e2e/policies.test-suite/balloons/n4c16/test01-basic-placement/code.var.sh

@@ -2,7 +2,7 @@
 # reserved and shared CPUs.
 
 cleanup() {
-    vm-command "kubectl delete pods -n kube-system pod0; kubectl delete pods -n three --all --now; kubectl delete pods --all --now; kubectl delete namespace three"
+    vm-command "kubectl delete pods -n kube-system pod0; kubectl delete pods --all --now --wait; kubectl delete namespace three --now --wait --ignore-not-found"
     return 0
 }
 

test/e2e/policies.test-suite/balloons/n4c16/test02-prometheus-metrics/code.var.sh

@@ -1,7 +1,7 @@
 # This test verifies prometheus metrics from the balloons policy.
 
 cleanup() {
-    vm-command "kubectl delete pods --all --now"
+    vm-command "kubectl delete pods --all --now --wait"
     return 0
 }
 
@@ -55,17 +55,17 @@ verify-metrics-has-line 'balloon_type="flex"'
 verify-metrics-has-line 'balloon_type="flex".* 5'
 
 # check deflating a balloon in metrics
-kubectl delete pods --now pod3
+kubectl delete pods --now --wait --ignore-not-found pod3
 verify-metrics-has-line 'balloon_type="flex"'
 verify-metrics-has-line 'balloon_type="flex".* 2'
 
-kubectl delete pods --now pod4
+kubectl delete pods --now --wait --ignore-not-found pod4
 sleep 5
 # check popping a balloon from metrics
 verify-metrics-has-no-line 'balloon_type="flex"'
 
 # pop fast-dualcore[0], keep fast-dualcore[1]
-kubectl delete pods --now pod1
+kubectl delete pods --now --wait --ignore-not-found pod1
 verify-metrics-has-line 'balloon="fast-dualcore\[1\]"'
 sleep 5
 verify-metrics-has-no-line 'balloon="fast-dualcore\[0\]"'

test/e2e/policies.test-suite/balloons/n4c16/test03-reserved/code.var.sh

@@ -3,15 +3,15 @@ cri_resmgr_cfg=${TEST_DIR}/balloons-reserved.cfg launch cri-resmgr
 
 cleanup() {
     vm-command \
-        "kubectl delete pod -n kube-system --now pod0
-         kubectl delete pod -n monitor-mypods --now pod1
-         kubectl delete pod -n system-logs --now pod2
-         kubectl delete pod -n kube-system --now pod3
-         kubectl delete pods --now pod4 pod5 pod6
-         kubectl delete pod -n kube-system --now pod7
-         kubectl delete namespace monitor-mypods
-         kubectl delete namespace system-logs
-         kubectl delete namespace my-exact-name"
+        "kubectl delete pod -n kube-system --now --wait --ignore-not-found pod0
+         kubectl delete pod -n monitor-mypods --now --wait --ignore-not-found pod1
+         kubectl delete pod -n system-logs --now --wait --ignore-not-found pod2
+         kubectl delete pod -n kube-system --now --wait --ignore-not-found pod3
+         kubectl delete pods --now --wait --ignore-not-found pod4 pod5 pod6
+         kubectl delete pod -n kube-system --now --wait --ignore-not-found pod7
+         kubectl delete namespace monitor-mypods --wait --ignore-not-found
+         kubectl delete namespace system-logs --wait --ignore-not-found
+         kubectl delete namespace my-exact-name --wait --ignore-not-found"
     return 0
 }
 

test/e2e/policies.test-suite/balloons/n4c16/test05-namespace/code.var.sh

@@ -3,15 +3,11 @@ cri_resmgr_cfg=${TEST_DIR}/balloons-namespace.cfg launch cri-resmgr
 
 cleanup() {
     vm-command \
-        "kubectl delete pods -n e2e-a --all --now
-         kubectl delete pods -n e2e-b --all --now
-         kubectl delete pods -n e2e-c --all --now
-         kubectl delete pods -n e2e-d --all --now
-         kubectl delete pods --all --now
-         kubectl delete namespace e2e-a
-         kubectl delete namespace e2e-b
-         kubectl delete namespace e2e-c
-         kubectl delete namespace e2e-d"
+        "kubectl delete pods --all --now --wait
+         kubectl delete namespace e2e-a --wait --ignore-not-found
+         kubectl delete namespace e2e-b --wait --ignore-not-found
+         kubectl delete namespace e2e-c --wait --ignore-not-found
+         kubectl delete namespace e2e-d --wait --ignore-not-found"
     return 0
 }
 cleanup

test/e2e/policies.test-suite/balloons/n4c16/test06-update-configmap/code.var.sh

@@ -4,11 +4,9 @@
 testns=e2e-balloons-test06
 
 cleanup() {
-    vm-command "kubectl delete pods --all --now; \
-        kubectl delete pods -n $testns --all --now; \
-        kubectl delete pods -n btype1ns0 --all --now; \
-        kubectl delete namespace $testns || :; \
-        kubectl delete namespace btype1ns0 || :"
+    vm-command "kubectl delete pods --all --now --wait; \
+        kubectl delete namespace $testns --now --wait --ignore-not-found || :; \
+        kubectl delete namespace btype1ns0 --now --wait --ignore-not-found || :"
     terminate cri-resmgr
     terminate cri-resmgr-agent
     vm-command "cri-resmgr -reset-policy; cri-resmgr -reset-config"

test/e2e/policies.test-suite/balloons/n4c16/test07-maxballoons/code.var.sh

@@ -1,5 +1,5 @@
 cleanup() {
-    vm-command "kubectl delete pods --all --now"
+    vm-command "kubectl delete pods --all --now --wait"
     return 0
 }
 
@@ -30,7 +30,7 @@ vm-command "kubectl describe pod pod2"
 if ! grep -q 'no suitable balloon instance available' <<< "$COMMAND_OUTPUT"; then
     error "could not find 'no suitable balloon instance available' in pod2 description"
 fi
-vm-command "kubectl delete pod pod2 --now"
+vm-command "kubectl delete pod pod2 --now --wait --ignore-not-found"
 
 # pod2: create dynamically the first dynamictwo balloon
 CPUREQ="800m" CPULIM="800m"
@@ -59,7 +59,7 @@ vm-command "kubectl describe pod pod5"
 if ! grep -q 'no suitable balloon instance available' <<< "$COMMAND_OUTPUT"; then
     error "could not find 'no suitable balloon instance available' in pod6 description"
 fi
-vm-command "kubectl delete pod pod5 --now"
+vm-command "kubectl delete pod pod5 --now --wait --ignore-not-found"
 
 cleanup
 

test/e2e/policies.test-suite/balloons/n4c16/test08-numa/code.var.sh

@@ -76,7 +76,7 @@ verify 'len(cpus["pod0c0"]) == 4' \
        'disjoint_sets(cpus["pod0c0"], cpus["pod6c0"])'
 
 # Leave only one guaranteed container to the first balloon.
-kubectl delete pods pod1 pod2 pod3 --now
+kubectl delete pods pod1 pod2 pod3 --now --wait --ignore-not-found
 report allowed
 verify 'len(cpus["pod0c0"]) == 1' \
        'len(cpus["pod4c0"]) == 1' \
@@ -87,7 +87,7 @@ verify 'len(cpus["pod0c0"]) == 1' \
 
 # Leave only bestefforts to the first balloon. Make sure they still
 # have a CPU.
-kubectl delete pods pod4 --now
+kubectl delete pods pod4 --now --wait --ignore-not-found
 report allowed
 verify 'len(cpus["pod0c0"]) == 1' \
        'len(cpus["pod5c0"]) == 1' \

test/e2e/policies.test-suite/balloons/n4c32/test01-dynamic-baloons/code.var.sh

@@ -42,7 +42,7 @@ verify-metrics-has-line 'sharedidlecpus_count="1"'
 verify 'len(nodes["pod8c0"])==2' \
        'len(dies["pod8c0"])==1' \
        'len(packages["pod8c0"])==1'
-kubectl delete pod pod8 --now
+kubectl delete pod pod8 --now --wait --ignore-not-found
 verify-metrics-has-no-line 'cpus_count="5"'
 
 # pod9: Add one more pod with 4 CPUs to inflate over dies, which should cross
@@ -54,7 +54,7 @@ create multicontainerpod
 verify 'len(nodes["pod9c0"])==4' \
        'len(dies["pod9c0"])==2' \
        'len(packages["pod9c0"])==1'
-kubectl delete pod pod9 --now
+kubectl delete pod pod9 --now --wait --ignore-not-found
 verify 'disjoint_sets(nodes["pod0c0"], nodes["pod1c0"], nodes["pod2c0"], nodes["pod3c0"], nodes["pod4c0"], nodes["pod5c0"], nodes["pod6c0"], nodes["pod7c0"])' \
 
 # pod9: Add one more pod with 7 CPUs to inflate over packages, which should cross
@@ -71,6 +71,6 @@ verify-metrics-has-no-line 'sharedidlecpus_count="1"'
 
 # pod0, pod9 deflate. This should free up 10 CPUs that will cause having
 # shared CPUs available again.
-kubectl delete pod pod10 --now
-kubectl delete pod pod0 --now
+kubectl delete pod pod10 --now --wait --ignore-not-found
+kubectl delete pod pod0 --now --wait --ignore-not-found
 verify-metrics-has-line 'sharedidlecpus_count="1"'

test/e2e/policies.test-suite/podpools/n4c16/test01-basic-placement/code.var.sh

@@ -1,7 +1,7 @@
 # Test placing containers with and without annotations to correct pools
 # reserved and shared CPUs.
 
-( kubectl delete pods pod3 -n kube-system --now ) || true
+( kubectl delete pods pod3 -n kube-system --now --wait --ignore-not-found ) || true
 
 # pod0: singlecpu
 out ""
@@ -42,7 +42,7 @@ verify 'cpus["pod3c0"] == cpus["pod3c1"] == cpus["pod3c2"]' \
        'cpus["pod3c0"] == expected.cpus.reserved[0]' \
        'mems["pod3c0"] == expected.mems.reserved[0]'
 
-kubectl delete pods pod3 -n kube-system --now
+kubectl delete pods pod3 -n kube-system --now --wait --ignore-not-found
 
 # pod4: bad pool name
 out ""
@@ -52,4 +52,4 @@ report allowed
 verify 'cpus["pod4c0"] == expected.cpus.default[0]' \
        'mems["pod4c0"] == expected.mems.default[0]'
 
-kubectl delete pods pod0 pod1 pod2 --now
+kubectl delete pods pod0 pod1 pod2 --now --wait --ignore-not-found

test/e2e/policies.test-suite/podpools/n4c16/test02-fill-order/code.var.sh

@@ -35,15 +35,15 @@ verify 'cpus["pod0c0"] == cpus["pod0c1"]' \
        'cpus["pod5c0"] == cpus["pod2c0"]' # the last pool should have been filled by pods 2 and 5
 
 # make a little room to the first pool and clear the last pool
-kubectl delete pods pod0 pod2 pod5 --now
+kubectl delete pods pod0 pod2 pod5 --now --wait --ignore-not-found
 
 # pod6: Balanced fill order should place this pod to the last pool (it has maximal free space)
 POD_ANNOTATION="pool.podpools.cri-resource-manager.intel.com: singlecpu" CONTCOUNT=1 create podpools-busybox
 report allowed
 verify 'disjoint_sets(cpus["pod6c0"],
                       set.union(cpus["pod1c0"], cpus["pod3c0"], cpus["pod4c0"]))'
 
-kubectl delete pods --all --now
+kubectl delete pods --all --now --wait
 reset counters
 
 out "### Filling dualcpu pool in Packed fill order"
@@ -60,7 +60,7 @@ verify 'cpus["pod0c0"] == cpus["pod1c0"] == cpus["pod2c0"]' \
        'disjoint_sets(cpus["pod0c0"], cpus["pod3c0"])'
 
 # Deleting two pods from the first pool, one from the last.
-kubectl delete pods pod0 pod1 pod5
+kubectl delete pods pod0 pod1 pod5 --now --wait --ignore-not-found
 
 # pod6: Packed fill order should place this to the last pool (it has minimal free space)
 POD_ANNOTATION="pool.podpools.cri-resource-manager.intel.com: dualcpu" CONTCOUNT=1 create podpools-busybox