Frequently Asked Questions¶

This is an index page for FAQs and troubleshooting for DCE 5.0.

Installation¶

• UI Login Issues
  • Cannot open DCE 5.0 interface after installation is complete; you can execute the diag.sh script for quick troubleshooting
  • DCE 5.0 login interface cannot be opened due to VIP access issues when installing with Metallb
• bootstrap node Issues
  • After shutting down and restarting the bootstrap node, the kind cluster cannot restart properly
  • Missing ip6tables when deploying Ubuntu 20.04 as a seed machine
  • After disabling IPv6 during installation, Podman on the bootstrap node cannot create containers
  • After restarting the bootstrap node's kind container, the kubelet service cannot start
  • How to uninstall data from the bootstrap node
• Certificate Issues
  • The kubeconfig for the global service cluster needs to be updated on the seed's replica
  • Certificate updates and kubeconfig for the kind cluster on the bootstrap node
  • After installing Contour, the default certificate validity period is only one year and does not auto-renew, causing the contour-envoy component to restart continuously after expiration
• Operating System Related Issues
  • Error during installation on CentOS 7.6
  • CentOS environment preparation issues
• Community Edition Installation Issues
  • Redis gets stuck during DCE 5.0 reinstallation in kind cluster
  • Failed to install fluent-bit in the community edition

Workbench¶

• Pipeline Related Issues
  • Error when executing the pipeline
  • How to update the podTemplate image for built-in Label?
  • When the pipeline build environment is Maven, how to modify the dependency package source in settings.xml?
  • When building images through Jenkins, containers cannot access private registry
  • How to modify the concurrency execution count of Jenkins pipelines
  • What to do if the pipeline running status does not update in time?
• GitOps Related Issues
  • Error when adding a GitHub repository under the GitOps module
  • When adding a repository in the GitOps module under a certain workspace, it prompts that the repository already exists
• Toolchain Related Issues
  • When the pipeline runs with Maven as the agent and uses integrated SonarQube for Java code scanning, it reports an error
  • What are the precautions when deploying GitLab

Container Management¶

• Permission issues in container management and global management modules
• Helm Installation:
  • Helm application installation fails with “OOMKilled” error
  • Unable to pull kpanda-shell image when installing application with Helm
  • Helm Chart interface does not display the latest Chart uploaded to Helm Repo
  • When Helm application installation fails, it gets stuck during installation and cannot delete the application to reinstall
• Workload -> After deleting node affinity and other scheduling policies, scheduling exceptions occur
• Application Backup:
  • What is the logic for Kcoral application backup to detect Velero status in the working cluster
  • When backing up and restoring across clusters, how does Kcoral obtain available clusters
  • Kcoral backed up Pods and Deployments with the same label, but after restoring the backup, two Pods appear
• Why do the corresponding elastic scaling records still exist after uninstalling VPA, HPA, and CronHPA
• Why does the console open abnormally in lower version clusters
• Creating and Integrating Clusters:
  • How to reset a created cluster
  • Failed to install plugins when integrating the cluster
  • When creating a cluster, enabling Kernel tuning for new clusters in advanced settings fails the cluster creation
  • After the cluster is disconnected, the kpanda-system namespace remains in Terminating state

MultiCloud Management¶

• The kernel for multicloud management is Karmada; what is the currently supported version of Karmada? Can a specific version be specified? Can it be upgraded?
• How to seamlessly migrate single-cluster applications to multicloud management?
• Does multicloud management support cross-cluster application log collection?
• For workloads distributed to multiple clusters in multicloud management, can monitoring information be presented in one view?
• Can multicloud management workloads communicate across clusters?
• Can multicloud management services achieve cross-cluster service discovery?
• Is there production-level support for multicloud management?
• How does multicloud management achieve failover?
• Multi-cluster permission issues
• How to query events from multiple clusters in multicloud management?
• After creating a multicloud application through multicloud management, how can relevant resource information be obtained through container management?
• How to customize the Karmada image source repository address in multicloud management?
• How to connect to multicloud clusters?
• Is it possible to delete only multicloud instances without deleting the components of multicloud management?
• How to achieve network intercommunication among multiple working clusters within a multicloud instance?

Cloud Native Networking¶

• kube-proxy Issues
  • IPVS Mode
  • iptables Mode
  • externalIPs not working under externalTrafficPolicy: Local
  • New endpoint rules for Service take a long time to take effect when updating endpoints
  • LoadBalancerSourceRanges not working properly under nftables mode
  • iptables nft and legacy mode selection issues
• Calico Issues
  • Offloading VXLAN causes latency
  • VXLAN parent device changed but routing not updated
  • Cluster calico-kube-controllers cache desynchronization causing memory leaks
  • Inter-node pod networking not working in IPIP mode
  • iptables nft and legacy mode selection issues
• Known Issues in Spiderpool v0.9
  • Error when SpiderCoordinator synchronizes status, but status remains running
  • Values.multus.multusCNI.uninstall setting ineffective, causing multus resources not to be deleted correctly
  • Unable to obtain serviceCIDR from kubeControllerManager Pod when kubeadm-config is missing
  • Upgrading from v0.7.0 to v0.9.0 causes panic due to new TxQueueLen property in SpiderCoordinator CRD
  • Due to different cluster deployment methods, SpiderCoordinator returns empty serviceCIDR, preventing Pod creation
• Known Issues in Spiderpool v0.8
  • ifacer cannot create bond using vlan 0
  • Disabling multus functionality still creates multus CR resources
  • SpiderCoordinator cannot detect gateway connections in Pod's netns
  • spiderpool-agent Pod crashes when kubevirt fixed IP feature is turned off
  • SpiderIPPool resources do not inherit gateway and route properties from SpiderSubnet
• Known Issues in Spiderpool v0.7
  • StatefulSet type Pods report IP conflict when obtaining IP allocation after restart
  • Spiderpool cannot recognize certain third-party controllers, causing Pods in StatefulSet to be unable to use fixed IP
  • Empty spidermultusconfig.spec causes spiderpool-controller Pod to crash
  • Cilium mode obtains incorrect overlayPodCIDR
  • In scenarios with a 1:1 Pod to IP ratio, IPAM allocation blockage occurs, preventing some Pods from running and affecting IP allocation performance
  • Disabling IP GC feature causes spiderpool-controller component to fail to start correctly due to readiness health check failure
  • IPPool.Spec.MultusName namespace/multusName resolution error causes associated multusName to be unfindable

Cloud Native Storage¶

• How does HwameiStor scheduler work in Kubernetes platform?
• How does HwameiStor handle scheduling for multi-replica workloads? How is it different from traditional general-purpose shared storage?
• How to operate and maintain a data volume on a Kubernetes node?
• How to handle errors when viewing LocalStorageNode?
• Why is StorageClass not automatically created after installing hwameistor-operator?

Virtual Machine¶

• API error on the virtual machine page
• Virtual machine creation failed
• Virtual machine created successfully but cannot be used
• VNC can start but network is inaccessible

Insight¶

• Clock skew in trace data
• Log collection troubleshooting guide
• Trace collection troubleshooting guide
• Using Insight to locate application anomalies
• What to do when ElasticSearch data is full?
• How to configure container log blacklist

Microservices Engine¶

• x-kubernets-validations error issue with skoala-init
• Nacos version downgrade issue

Service Mesh¶

• Cannot find the associated cluster when creating a mesh
• Mesh creation is stuck in "Creating" and ultimately fails
• Created mesh is abnormal but cannot be deleted
• Managed mesh management cluster failed
• Anomalies with istio-ingressgateway when managing mesh management cluster
• Mesh space cannot unbind properly
• Tracking issues to integrate DCE 4.0
• Sidecar configuration conflicts with workload sidecar
• Multi-cloud interconnect anomalies in managed mesh
• Sidecar occupies a lot of memory
• When creating a mesh, the cluster list contains unknown clusters
• Managed mesh APIServer certificate expiration handling methods
• Common 503 errors in service mesh
• How to allow applications listening on localhost in the cluster to be accessed by other Pods

Middleware¶

• MySQL Troubleshooting
  • MySQL health check
  • MySQL Pod issues
  • MySQL Operator issues
  • MySQL master-slave relationship issues
  • MySQL MGR parameter configuration issues
  • MySQL MGR troubleshooting manual
  • Other MySQL failure cases
  • MySQL master-slave mode response to network interruptions
• Elasticsearch Troubleshooting
  • Elasticsearch PVC disk capacity full
  • Elasticsearch business index alias is occupied
  • Error setting GoMAXPROCS for operator
  • Error terminating due to java.lang.OutOfMemoryError: Java heap space
  • Error during Elasticsearch installation in OCP environment: Operation not permitted
  • One node has abnormal disk read throughput and high CPU workload
  • Error status:429 when writing data to Elasticsearch

AI Lab¶

• Cluster not found in AI Lab dropdown list
• AI Lab Notebook not controlled by queue quotas
• AI Lab queue initialization failed

Global Management¶

• Why can't istio-ingressgateway start after restarting the cluster (virtual machine)?
• Login infinite loop, error 401 or 403
• Keycloak cannot start
• Upgrade fails when upgrading global management alone

Permission Issues¶

• Container management permission description
• Microservices engine permission description
• Application workbench permission description
• Service mesh permission description
• Middleware permission description
• AI Lab permission description

Comments