Loading...

XML

Word

Printable

Type: Bug
Resolution: Done-Errata
Priority: Critical
Fix Version/s: 4.15.0
Affects Version/s: 4.12.z
Component/s: Networking / ovn-kubernetes
Labels:
- OVN-Kubernetes
- TestBlocker
- Triaged
- azure
- ipi

Severity:
Moderate
Regression:
No
Sprint:
SDN Sprint 234, SDN Sprint 235, SDN Sprint 236, SDN Sprint 237, SDN Sprint 238, SDN Sprint 239, SDN Sprint 240, SDN Sprint 241
sprint_count:
8
Release Blocker:
Approved
Blocked:
False
Blocked Reason:

Hide

SNO installation on 4.14 fails in Azure/GCP

Show
SNO installation on 4.14 fails in Azure/GCP
Target Version:

4.14.0

SFDC Cases Counter:
SFDC Cases Open:
SFDC Cases Links:

PX Impact Score:
PX Priority Data:
PX Review Complete:

Description of problem:

OpenShift Container Platform 4.12.5 installation with IPI installation method on Microsoft Azure is showing undesired behavior when trying to curl "https://api.<clustername>.<domain>:6443/readyz". When using `HostNetwork` it all works without any issues. But when doing the same request from a pod that does not have `HostNetwork` capabilties and therefore has an IP from the SDN range, a big portion of the requests is failing.

$ oc get clusterversion
NAME      VERSION   AVAILABLE   PROGRESSING   SINCE   STATUS
version   4.12.5    True        False         29m     Cluster version is 4.12.5

$ oc get network cluster -o yaml
apiVersion: config.openshift.io/v1
kind: Network
metadata:
  creationTimestamp: "2023-03-10T13:12:06Z"
  generation: 2
  name: cluster
  resourceVersion: "2975"
  uid: e1e9c464-526c-4ebf-ab84-0deedf092cac
spec:
  clusterNetwork:
  - cidr: 10.128.0.0/14
    hostPrefix: 23
  externalIP:
    policy: {}
  networkType: OVNKubernetes
  serviceNetwork:
  - 172.30.0.0/16
status:
  clusterNetwork:
  - cidr: 10.128.0.0/14
    hostPrefix: 23
  clusterNetworkMTU: 1400
  networkType: OVNKubernetes
  serviceNetwork:
  - 172.30.0.0/16

$ oc get infrastructure cluster -o yaml
apiVersion: config.openshift.io/v1
kind: Infrastructure
metadata:
  creationTimestamp: "2023-03-10T13:12:04Z"
  generation: 1
  name: cluster
  resourceVersion: "430"
  uid: 5c260276-d901-40f7-a28c-172c492e81e6
spec:
  cloudConfig:
    key: config
    name: cloud-provider-config
  platformSpec:
    type: Azure
status:
  apiServerInternalURI: https://api-int.clustername.domain.lab:6443
  apiServerURL: https://api.clustername.domain.lab:6443
  controlPlaneTopology: HighlyAvailable
  etcdDiscoveryDomain: ""
  infrastructureName: sreberazure-njj24
  infrastructureTopology: HighlyAvailable
  platform: Azure
  platformStatus:
    azure:
      cloudName: AzurePublicCloud
      networkResourceGroupName: sreberazure-njj24-rg
      resourceGroupName: sreberazure-njj24-rg
    type: Azure

$ oc project openshift-apiserver
Already on project "openshift-apiserver" on server "https://api.clustername.domain.lab:6443".
$ oc get pod
NAME                         READY   STATUS    RESTARTS   AGE
apiserver-6f58784797-kq4kr   2/2     Running   0          41m
apiserver-6f58784797-l69jr   2/2     Running   0          38m
apiserver-6f58784797-nn6tn   2/2     Running   0          45m

$ oc get pod -o wide
NAME                         READY   STATUS    RESTARTS   AGE   IP            NODE                         NOMINATED NODE   READINESS GATES
apiserver-6f58784797-kq4kr   2/2     Running   0          42m   10.130.0.21   sreberazure-njj24-master-0   <none>           <none>
apiserver-6f58784797-l69jr   2/2     Running   0          38m   10.129.0.29   sreberazure-njj24-master-2   <none>           <none>
apiserver-6f58784797-nn6tn   2/2     Running   0          45m   10.128.0.36   sreberazure-njj24-master-1   <none>           <none>

$ oc rsh apiserver-6f58784797-l69jr
Defaulted container "openshift-apiserver" out of: openshift-apiserver, openshift-apiserver-check-endpoints, fix-audit-permissions (init)
sh-4.4# while true; do curl -k --connect-timeout 1  https://api.clustername.domain.lab:6443/readyz; sleep 1; done
curl: (28) Connection timed out after 1000 milliseconds
okokokcurl: (28) Connection timed out after 1001 milliseconds
okokcurl: (28) Connection timed out after 1003 milliseconds
curl: (28) Connection timed out after 1001 milliseconds
curl: (28) Connection timed out after 1001 milliseconds
okokokokokokokokokcurl: (28) Connection timed out after 1001 milliseconds
okokcurl: (28) Connection timed out after 1001 milliseconds
curl: (28) Connection timed out after 1001 milliseconds
^C
sh-4.4# exit
exit
command terminated with exit code 130

$ oc project openshift-kube-apiserver
Now using project "openshift-kube-apiserver" on server "https://api.clustername.domain.lab:6443".

$ oc get pod -o wide
NAME                                              READY   STATUS      RESTARTS   AGE   IP            NODE                         NOMINATED NODE   READINESS GATES
apiserver-watcher-sreberazure-njj24-master-0      1/1     Running     0          55m   10.0.0.6      sreberazure-njj24-master-0   <none>           <none>
apiserver-watcher-sreberazure-njj24-master-1      1/1     Running     0          57m   10.0.0.8      sreberazure-njj24-master-1   <none>           <none>
apiserver-watcher-sreberazure-njj24-master-2      1/1     Running     0          57m   10.0.0.7      sreberazure-njj24-master-2   <none>           <none>
installer-2-sreberazure-njj24-master-2            0/1     Completed   0          51m   10.129.0.27   sreberazure-njj24-master-2   <none>           <none>
installer-3-sreberazure-njj24-master-2            0/1     Completed   0          50m   10.129.0.32   sreberazure-njj24-master-2   <none>           <none>
installer-4-sreberazure-njj24-master-2            0/1     Completed   0          49m   10.129.0.36   sreberazure-njj24-master-2   <none>           <none>
installer-5-sreberazure-njj24-master-2            0/1     Completed   0          46m   10.129.0.15   sreberazure-njj24-master-2   <none>           <none>
installer-6-sreberazure-njj24-master-0            0/1     Completed   0          37m   10.130.0.27   sreberazure-njj24-master-0   <none>           <none>
installer-6-sreberazure-njj24-master-1            0/1     Completed   0          39m   10.128.0.45   sreberazure-njj24-master-1   <none>           <none>
installer-6-sreberazure-njj24-master-2            0/1     Completed   0          36m   10.129.0.37   sreberazure-njj24-master-2   <none>           <none>
kube-apiserver-guard-sreberazure-njj24-master-0   1/1     Running     0          37m   10.130.0.29   sreberazure-njj24-master-0   <none>           <none>
kube-apiserver-guard-sreberazure-njj24-master-1   1/1     Running     0          38m   10.128.0.47   sreberazure-njj24-master-1   <none>           <none>
kube-apiserver-guard-sreberazure-njj24-master-2   1/1     Running     0          50m   10.129.0.31   sreberazure-njj24-master-2   <none>           <none>
kube-apiserver-sreberazure-njj24-master-0         5/5     Running     0          37m   10.0.0.6      sreberazure-njj24-master-0   <none>           <none>
kube-apiserver-sreberazure-njj24-master-1         5/5     Running     0          38m   10.0.0.8      sreberazure-njj24-master-1   <none>           <none>
kube-apiserver-sreberazure-njj24-master-2         5/5     Running     0          34m   10.0.0.7      sreberazure-njj24-master-2   <none>           <none>
revision-pruner-6-sreberazure-njj24-master-0      0/1     Completed   0          33m   10.130.0.35   sreberazure-njj24-master-0   <none>           <none>
revision-pruner-6-sreberazure-njj24-master-1      0/1     Completed   0          33m   10.128.0.56   sreberazure-njj24-master-1   <none>           <none>
revision-pruner-6-sreberazure-njj24-master-2      0/1     Completed   0          33m   10.129.0.39   sreberazure-njj24-master-2   <none>           <none>

$ oc rsh kube-apiserver-sreberazure-njj24-master-1
sh-4.4# while true; do curl -k --connect-timeout 1  https://api.clustername.domain.lab:6443/readyz; sleep 1; done
okokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokok

Also changing  `--connect-timeout 1` from curl to `--connect-timeout 10` for example does not have any impact. It simply takes longer until the timeout is reached.

Version-Release number of selected component (if applicable):

OpenShift Container Platform 4.12 (also previous version were not tested)

How reproducible:

Always

Steps to Reproduce:

1. Install OpenShift Container Platform 4.12 on Azure using IPI install method and set the SDN to OVN-Kubernetes
2. Once successfully installed run `oc project openshift-apiserver`
3. rsh apiserver-<podID>
4. while true; do curl -k --connect-timeout 1  https://api.clustername.domain.lab:6443/readyz; sleep 1; done

Actual results:

sh-4.4# while true; do curl -k --connect-timeout 1  https://api.clustername.domain.lab:6443/readyz; sleep 1; done
curl: (28) Connection timed out after 1000 milliseconds
okokokcurl: (28) Connection timed out after 1001 milliseconds
okokcurl: (28) Connection timed out after 1003 milliseconds
curl: (28) Connection timed out after 1001 milliseconds
curl: (28) Connection timed out after 1001 milliseconds
okokokokokokokokokcurl: (28) Connection timed out after 1001 milliseconds
okokcurl: (28) Connection timed out after 1001 milliseconds
curl: (28) Connection timed out after 1001 milliseconds

Expected results:

sh-4.4# while true; do curl -k --connect-timeout 1  https://api.clustername.domain.lab:6443/readyz; sleep 1; done
okokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokokok

Additional info:

blocks

OCPBUGS-18159 Azure; NLB; OVN-K: Requests from CNI pods to internalAPI server domain fails always

Closed

links to

openshift/machine-config-operator#3878: OCPBUGS-9972: Fix azure routes hack for ovnk pods towards internalLB on master nodes in SGW mode

RHSA-2023:5006 OpenShift Container Platform 4.14.z security update

Assignee:: Surya Seetharaman

Reporter:: Simon Reber

QA Contact:: Huiran Wang

Votes:: 0 Vote for this issue

Watchers:: 9 Start watching this issue

Created:: 2023/03/10 2:33 PM

Updated:: 2023/10/31 1:43 PM

Resolved:: 2023/10/31 1:43 PM

Details

Description

Attachments

Issue Links

Easy Agile Planning Poker

Activity

People

Dates