add u2o test case #2203
add u2o test case #2203
Conversation
changluyi
force-pushed
the
u2o_test_case
branch
2 times, most recently
from
b879dcf
to
79911d9
Compare
| restartCmd := "kubectl rollout restart deployment kube-ovn-controller -n kube-system" | ||
| _, err = exec.Command("bash", "-c", restartCmd).CombinedOutput() |
There was a problem hiding this comment.
Use the golang client to restart the deployment.
| framework.ExpectEqual(int(subnet.Status.V4UsingIPs), 1) | ||
| } | ||
|
|
||
| agStr := strings.Replace(fmt.Sprintf("%s.u2o_exclude_ip.ip4", subnet.Name), "-", ".", -1) |
There was a problem hiding this comment.
| agStr := strings.Replace(fmt.Sprintf("%s.u2o_exclude_ip.ip4", subnet.Name), "-", ".", -1) | |
| asName := strings.Replace(fmt.Sprintf("%s.u2o_exclude_ip.ip4", subnet.Name), "-", ".", -1) |
changluyi
force-pushed
the
u2o_test_case
branch
from
79911d9
to
36df155
Compare
| modifiedSubnet := subnet.DeepCopy() | ||
| modifiedSubnet.Spec.U2OInterconnection = false | ||
| subnetClient.PatchSync(subnet, modifiedSubnet) | ||
| time.Sleep(5 * time.Second) |
There was a problem hiding this comment.
Why sleep 5s here? To wait for the subnet to be ready?
There was a problem hiding this comment.
yes， if create a pod and subnet adding or updating at the same time, it will trigger IP count error sometimes. and This problem seems to have always existed
There was a problem hiding this comment.
We should wait until the subnet is ready instead of sleeping a period.
There was a problem hiding this comment.
yes， if create a pod and subnet adding or updating at the same time, it will trigger IP count error sometimes. and This problem seems to have always existed
I think this is a bug. Sometimes a user creates subnet(s) and pod(s) using a single YAML file.
There was a problem hiding this comment.
sorry, I describe it wrong, it causes errors only in updating the subnet, but not creating the subnet.
because when you update the subnet, the subnet status is always ready.
There was a problem hiding this comment.
How about waiting for .status.u2OInterconnectionIP to be changed?
There was a problem hiding this comment.
it may not work, status.u2oInterconnectionIP is changed after function reconcileU2OInterconnectionIP
and if there is a pod added after the time subnet executes reconcileU2OInterconnectionIP and before the subnet update function handleAddOrUpdateSubnet ends， the subnet usingIps count may not include this pod.
I originally used sleep to do a workaround, because once the u2o switch is switched, all pods have to be rebuilt
| v4UPodIP, v6UPodIP := util.SplitStringIP(underlayPod.Status.PodIP) | ||
| v4OPodIP, v6OPodIP := util.SplitStringIP(overlayPod.Status.PodIP) |
| @@ -22,6 +25,7 @@ import ( | |||
| ) | |||
|
|
|||
| const dockerNetworkName = "kube-ovn-vlan" | |||
| const curlListenPort = 8081 | |||
There was a problem hiding this comment.
It's OK to use a fixed port, but it would be better to use a random one.
| output, _ := exec.Command("bash", "-c", pingCmd).CombinedOutput() | ||
| ginkgo.By("checking curl reachable") | ||
| cmd := fmt.Sprintf("kubectl exec %s -n %s -- curl -q -s --connect-timeout 5 %s/clientip", podName, podNamespace, net.JoinHostPort(targetIP, targetPort)) | ||
| output, _ := exec.Command("bash", "-c", cmd).CombinedOutput() |
| isDefaultRouteExist := false | ||
| for _, route := range routes { | ||
| if route.Dst == "default" { | ||
| if util.CheckProtocol(route.Gateway) == apiv1.ProtocolIPv4 { | ||
| if isEnableU2O { | ||
| framework.ExpectEqual(route.Gateway, v4InterconnIp) | ||
| } else { | ||
| framework.ExpectEqual(route.Gateway, v4gw) | ||
| } | ||
| } else { | ||
| if isEnableU2O { | ||
| framework.ExpectEqual(route.Gateway, v6InterconnIp) | ||
| } else { | ||
| framework.ExpectEqual(route.Gateway, v6gw) | ||
| } | ||
| } | ||
| isDefaultRouteExist = true | ||
| } | ||
| } |
There was a problem hiding this comment.
We should check BOTH IPv4 and IPv6 for a dual-stack pod.
What type of this PR
Examples of user facing changes:
Which issue(s) this PR fixes:
Fixes #2050