Enable/Disable collection of audit logs¶
- Kubernetes Audit Logs: Kubernetes itself generates audit logs. When this feature is enabled, audit log files for Kubernetes will be created in the specified directory.
- Collecting Kubernetes Audit Logs: The log files mentioned above are collected using the Insight Agent. The prerequisite for collecting Kubernetes audit logs are that the cluster has enabled Kubernetes audit logs, the export of audit logs has been allowed, and the collection of audit logs has been opened.
AI platform Installation Status¶
- For AI Community installations, the Kubernetes audit log switch was not operated during the management cluster installation process.
- For AI platform installations, the Kubernetes audit log switch is enabled by default.
- To set it to default off, you can modify the installer's clusterConfig.yaml file (set logPath to empty "").
- The collection of Kubernetes audit logs switch is disabled by default for the management cluster.
- Default settings do not support configuration.
Management Cluster Collection of Kubernetes Audit Logs Switch¶
AI platform Installation Environment¶
Confirm Enabling Kubernetes Audit Logs¶
Run the following command to check if audit logs are generated under the /var/log/kubernetes/audit directory. If they exist, it means that Kubernetes audit logs are successfully enabled.
If they are not enabled, refer to the documentation on enabling/disabling Kubernetes audit logs.
Enable Collection of Kubernetes Audit Logs Process¶
-
Add ChartMuseum to the helm repo.
Modify the IP address in this command to the IP address of the Spark node.
Note
If using a self-built Harbor repository, please modify the chart repo URL in the first step to the insight-agent chart URL of the self-built repository.
-
Save the current Insight Agent helm values.
-
Get the current version number ${insight_version_code}.
-
Update the helm value configuration.
-
Restart all fluentBit pods under the insight-system namespace.
Disable Collection of Kubernetes Audit Logs¶
The remaining steps are the same as enabling the collection of Kubernetes audit logs, with only a modification in the previous section's step 4: updating the helm value configuration.
helm upgrade --install --create-namespace --version ${insight_version_code} --cleanup-on-fail insight-agent chartmuseum/insight-agent -n insight-system -f insight-agent-values-bak.yaml --set global.exporters.auditLog.kubeAudit.enabled=false
AI Community Online Installation Environment¶
Note
If installing AI Community in a Kind cluster, perform the following steps inside the Kind container.
Confirm Enabling Kubernetes Audit Logs¶
Run the following command to check if audit logs are generated under the /var/log/kubernetes/audit directory. If they exist, it means that Kubernetes audit logs are successfully enabled.
If they are not enabled, refer to the documentation on enabling/disabling Kubernetes audit logs.
Enable Collection of Kubernetes Audit Logs Process¶
-
Save the current values.
-
Get the current version number ${insight_version_code} and update the configuration.
-
Update the helm value configuration.
helm upgrade --install --create-namespace --version ${insight_version_code} --cleanup-on-fail insight-agent insight-release/insight-agent -n insight-system -f insight-agent-values-bak.yaml --set global.exporters.auditLog.kubeAudit.enabled=true
If the upgrade fails due to an unsupported version, check if the helm repo used in the command has that version. If not, retry after you updated the helm repo.
-
Restart all fluentBit pods under the insight-system namespace.
Disable Collection of Kubernetes Audit Logs¶
The remaining steps are the same as enabling the collection of Kubernetes audit logs, with only a modification in the previous section's step 3: updating the helm value configuration.
helm upgrade --install --create-namespace --version ${insight_version_code} --cleanup-on-fail insight-agent insight-release/insight-agent -n insight-system -f insight-agent-values-bak.yaml --set global.exporters.auditLog.kubeAudit.enabled=false
Change Worker Cluster¶
Each worker cluster is independent and can be turned on as needed.
Steps to Enable Audit Log Collection When Creating a Cluster¶
By default, the collection of K8s audit logs is turned off. If you need to enable it, you can follow these steps:
Set the switch to the enabled state to enable the collection of K8s audit logs.
When creating a worker cluster via AI platform, ensure that the K8s audit log option for the cluster is set to 'true' so that the created worker cluster will have audit logs enabled.
After the cluster creation is successful, the K8s audit logs for that worker cluster will be collected.
Steps to Enable/Disable After Accessing or Creating the Cluster¶
Confirm Enabling K8s Audit Logs¶
Run the following command to check if audit logs are generated under the /var/log/kubernetes/audit directory. If they exist, it means that K8s audit logs are successfully enabled.
If they are not enabled, refer to the documentation on enabling/disabling K8s audit logs.
Enable Collection of K8s Audit Logs¶
The collection of K8s audit logs is disabled by default. To enable it, follow these steps:
-
Select the cluster that has been accessed and needs to enable the collection of K8s audit logs.
-
Go to the Helm App management page and update the insight-agent configuration (if insight-agent is not installed, you can install it).
-
Enable/Disable the collection of K8s audit logs switch.
-
After enabling/disabling the switch, the fluent-bit pod needs to be restarted for the changes to take effect.