If Event Streams resources are not available, the following are possible sypmtoms and causes.
Event Streams not available after installation
After a successsful installation message is displayed, Event Streams might not be available to use yet.
It can take up to 10 minutes before Event Streams is available to use. The IBM Cloud Private installation might return a successful completion message before all Event Streams services start up.
If the installation continues to be unavailable, run the installation diognastics scripts.
Insufficient system resources
You can specify the memory and CPU requirements when Event Streams is installed. If the values set are larger than the resources available, then pods will fail to start.
Common error messages in such cases include the following:
pod has unbound PersistentVolumeClaims: occurs when there are no Persistent Volumes available that meet the requirements provided at the time of installation.Insufficient memory: occurs when there are no nodes with enough available memory to support the limits provided at the time of installation.Insufficient CPU: occurs when there are no nodes with enough available CPU to support the limits provided at the time of installation.
For example, if each Kafka broker is set to require 80 GB of memory on a system that only has 16 GB available per node, you might see the following error message:
To get detailed information on the cause of the error, check the events for the individual pods (not the logs at the stateful set level).
If a system has 16 GB of memory available per node, then the broker memory requirements must be set to be less than 16 GB. This allows resources to be available for the other Event Streams components which may reside on the same node.
To correct this issue, uninstall Event Streams. Install again using lower resource requirements, or increase the amount of system resources available to the pod.
Problems with secrets
When using a non-default Docker registry, you might need to provide a secret which stores the user ID and password to access that registry.
If there are issues with the secret that holds the user ID and password used to access the Docker registry, the events for a pod will show an error similar to the following.
To resolve this issue correct the secret and install Event Streams again.
Installation failure stating object already exists
If a secret that does not exist is specified during installation, the process fails even if no secret is required to access the Docker registry.
The default Docker image registry at ibmcom does not require a secret specifying the user ID and password.
To correct this, install Event Streams again without specifying a secret.
If you are using a Docker image registry that does require a secret, attempting to install again might fail stating that an object already exists, for example:
Internal service error : rpc error: code = Unknown desc = rolebindings.rbac.authorization.k8s.io "elh-ibm-es-secret-copy-crb-sys" already exists
Delete the left over object cited and other objects before trying to install again. For instructions, see how to fully clean up after uninstallation.