Known issues

This section lists known issues that you might run into while using the CDP Private Cloud Management Console service.

Inability to collect diagnostics bundle in an air gapped CDP Private Cloud Experiences deployment using ECS

Problem: In an air gapped CDP Private Cloud Experiences deployment using ECS, attempts to gather diagnostics data by using the Collect and Send Diagnostics option might fail when selecting either No Time Limit or Recent Time Range as the duration. The issue arises because of the host's system clock being out of sync with its NTP servers.

Workaround: If the diagnostics data collection fails with either No Time Limit or Recent Time Range as the duration, attempt to collect the data by selecting Custom Time Range as the duration, and providing valid start date and end date.

The Environments page on the Management Console UI for an environment in a deployment using ECS does not display the platform name

Problem: When you view the details of an environment using the Management Console UI in a CDP Private Cloud Experiences deployment using ECS, the Platform field appears blank.

Workaround: Use the relevant CDP CLI command from the environments module to view the required details.

The error message indicating an invalid storage class name when registering an environment pops up only once

Problem: When you try registering an environment multiple times after specifying an invalid storage class name, the corresponding error message popup window displays only after the first attempt and not after every attempt to register the environment.

Workaround:
  • Validate the storage class name that you specify when registering the environment. This is also part of the pre-installation checklist.
  • If you do not see the error message popup window or if the environment registration fails, then validate the class name on the OpenShift container deployment.

Environment registration is possible without specifying a storage class name or without configuring a default storage class

Problem: Management Console allows you to register an environment without providing a storage class, even if no default storage class is configured on the OpenShift cluster. Therefore, the scheduling does not work on the pods when the environment is created.

Workaround: When registering an environment, you must ensure that you have either provided a valid storage class, or that a default storage class is configured on the OpenShift cluster. Otherwise, you must register the environment again by providing the correct configuration details.

Cannot delete an environment when the registration of the compute cluster fails

Problem: If the CA certificates that you upload for the external database or vault are incorrect, when you attempt to register an environment, the registration succeeds but the corresponding compute cluster is not created. Deleting the environment then fails with the message "Compute cluster deregistration failed with error."

Workaround: Upload the correct CA certificates from the Administration tab of the Management Console, and then delete the environment.

Management Console allows registration of two environments of the same name

Problem: If two users attempt to register environments of the same name at the same time, this might result in an unusable environment.

Workaround: Delete the environment and ensure that only one user attempts to register a new environment.

Kerberos service does not always handle Cloudera Manager downtime

Problem: The Cloudera Manager Server in the base cluster must be running to generate Kerberos principals for CDP Private Cloud. If there is downtime, you might observe Kerberos-related errors.

Workaround: Resolve downtime issues on Cloudera Manager. If you encounter Kerberos errors, you can retry the concerned operation such as creating Virtual Warehouses.

The Test PromQL Expression option returns a generic unexpected error page when the PromQL expression is invalid

Problem: When configuring an alert rule, if you specify an invalid PromQL expression and click Test PromQL Expression, an unexpected error appears. In addition, all the information specified for configuring the alert rule is lost.

Workaround: Save the rule and then test the PromQL expression.

Issue with the Duplicate option for SNMP alert receivers

Problem: Attempting to duplicate an alert receiver of type SNMP results in alert configuration errors.

Workaround: Do not use the Duplicate option with an SNMP alert receiver.

Plain HTML available as tooltip text on the Management Console dashboard

Problem: In a few situations, the tooltip text on the Management Console dashboard displays plain HTML as the tooltip text.

Workaround: There is currently no workaround for the issue.

The Raw API Response window on the Management Console dashboard disappears after opening

Problem: On the Management Console dashboard, when you click to access the Raw API Response, the window disappears for a few seconds after opening.

Workaround: There is currently no workaround for the issue.

The Management Console dashboard is blank if Prometheus is inaccessible

Problem: If the control plane monitoring app cannot access Prometheus on the control plane, the Dashboard page that loads is blank. This also increases the average page load time.

Workaround: There is currently no workaround for the issue.

Alerts do not display on the Management Console dashboard in certain situations

Problem: When adding or modifying an alert rule, if aggregation operators are used in the PromQL expressions such that the result of the expression does not contain the mandatory labels, then the particular alert, when triggered, does not display on the Management Console.

Workaround: Metrics reported by the environments always contain the following labels: appId and appName. Therefore, the result of the alert rule's query expression also must contain these labels. To ensure that the result contains the labels, include the by (appId, appName) clause when using aggregation operators in the query expression. For example, instead of the count(my_metric) > 0 expression, use the count(my_metric) by (appId, appName) > 0 expression.

Management Console dashboard displays incorrect service values for certain alerts

Problem: For a triggered alert, if its the source is either the snmp-notifier or logger-alert-receiver monitoring pod, the Management Console dashboard displays the service value for that alert as Miscellaneous instead of Monitoring.

Workaround: There is currently no workaround for the issue.

Search results for alert rules are reset after disabling a rule

Problem: For an alert rule that appears as the result of a search on the Alerts page, enabling or disabling the alert rule from its quick links menu and saving the rule resets the search results.

Workaround: There is currently no workaround for the issue.

Alert rule filters are reset after editing a rule

Problem: Creating or editing an alert rule based on filter criteria and returning to the Alerts page resets the filters.

Workaround: There is currently no workaround for the issue.

The alert-admin logs are flooded with error messages during control plane upgrade

Problem: When upgrading the control plane services, the alert-admin logs are flooded with error messages.

Workaround: There is currently no workaround for the issue.

SDKs are not supported in the current release

Problem: SDKs are not supported in the current CDP Private Cloud release.

Workaround: You can use the Management Console user interface or the CLI commands to perform your tasks.