This release addresses a memory leak issue in the Kublr Agents caused by client-go v0.26.0.
We have reported this bug on Kuberntest under the issue Using Dial function with client-go kubernetes.NewForConfig triggers a memory leak, and the release agents are now updated to use client-go v0.25.4.
The Kublr team strongly recommends that all customers migrate to v1.25.3 and use the agents provided in this release.
| Version | Kublr Agent | Notes |
|---|---|---|
| 1.25 | 1.25.6-6 | Default version: v1.25.6 |
| 1.24 | 1.24.9-6 | |
| 1.23 | 1.23.15-6 | |
| 1.22 | 1.22.17-8 | Deprecated in 1.26.0 |
| 1.21 | 1.21.14-24 | End of support in 1.26.0 |
To quickly get started with Kublr, run the following command in your terminal:
sudo docker run --name kublr -d --restart=unless-stopped -p 9080:9080 kublr/kublr:1.25.3
The Kublr Demo/Installer docker container can be run on ARM-based PC, such as MacBook M1.
Follow the full instructions in Quick start for Kublr Demo/Installer.
To use Kublr in an airgap environment, you will need to download the following BASH scripts from the repository at https://repo.kublr.com:
You will also need to download the following Helm package archives and Docker images list:
| Component | Version | Kublr Agent | Notes |
|---|---|---|---|
| Kubernetes | 1.25 | 1.25.6-4 | default v1.25.6 |
| 1.24 | 1.24.9-4 | ||
| 1.23 | 1.23.15-4 | ||
| 1.22 | 1.22.17-4 | Deprecated in 1.26.0 | |
| 1.21 | 1.21.14-22 | End of support in 1.26.0 |
| Component | Version |
|---|---|
| Kublr Operator | 1.25.3 |
| Kublr Control Plane | 1.25.3 |
| Component | Version |
|---|---|
| Kubernetes | |
| Dashboard | v2.7.0 |
| Kublr System | 1.25.3 |
| LocalPath Provisioner (helm chart version) | 0.0.23-12 |
| Ingress | 1.25.3 |
| nginx ingress controller (helm chart version) | 4.2.3 |
| cert-manager (helm chart version) | 1.10.0 |
| Centralized Logging | 1.25.3 |
| ElasticSearch | 7.10.2 |
| SearchGuard | 53.6.0 |
| Kibana | 7.10.2 |
| SearchGuard Kibana plugin | 53.0.0 |
| SearchGuard Admin | 7.10.2-53.6.0 |
| OpenSearch (helm chart version) | 2.6.2 |
| OpenSearch Dashboards | (helm chart version) |
| RabbitMQ | 3.9.5 |
| Curator | 5.8.1 |
| Logstash | 7.10.2 |
| Fluentd | 4.3.3 |
| Fluentbit | 1.8.12 |
| Centralized Monitoring | 1.25.3 |
| Prometheus | 2.37.1 LTS |
| Kube State Metrics (helm chart version) | 4.24.0 |
| AlertManager | 0.22.0 |
| Grafana | 8.5.13 |
| Victoria Metrics | |
| Cluster | 0.9.59 |
| Agent | 0.8.35 |
| Alert | 0.5.21 |
For Kublr Kubernetes clusters deployed on vSphere, you may encounter an update error that requires manual unmounting of CD/DVD drive 1 from each virtual machine using the vCenter console.
Elasticsearch is only supported on the AMD64 architecture. If you are using an ARM64 cluster, you can use tag 7.17.9 for the cluster nodes.
vSphere CSI limitation: Please refer to the VMWare documentation for information on vSphere functionality supported by the vSphere Container Storage Plug-in.
CRI change limitation: When deploying an Ingress controller in hostPort mode (typically for vSphere or BareMetal installations), DNAT rules may not be cleaned properly. This is an open issue in the Felix project. If your Ingress NGINX controller connection times out, please reboot the worker nodes.
RedHat-based clusters may encounter installation problems with Docker. In such cases, please use the ContainerD runtime configuration:
spec:
kublrAgentConfig:
kublr:
setup:
runtime_fallback_order: containerd