Update a Managed Kafka cluster
Changes a cluster’s name, allowlist, broker storage, storage autoscaling, instance type, broker count, or tiered storage. Send only the fields to change. Every field is checked before anything is saved, so an invalid field leaves the cluster unchanged. A field already at the requested value is left as it is, so retrying after an earlier field was saved still applies the rest. allowedCIDRs, storageGiB, instanceType, brokers, and tieredStorage are then applied one after another, in that order; name and storageAutoscale are saved first and don’t apply to the host. Name also works on a BYOC cluster. Returns 409 while a previous change is still being applied. Company admins only.
Authorizations
API key passed as a Bearer token in the Authorization header.
Path Parameters
Body
Client IPv4 ranges allowed to reach the cluster. Replaces the current list. Empty allows any IP; clients still need the password. No broker restarts.
New number of brokers. Brokers can only be added. The count the cluster already has is left as it is. Existing partitions stay where they are; new topics use every broker.
3 <= x <= 30EC2 instance type for every broker. The brokers restart one at a time onto new machines. The type they already run is left as it is.
m8g.2xlarge, m8g.4xlarge, m8g.xlarge, m8i.2xlarge, m8i.4xlarge, m8i.xlarge, r8g.2xlarge, r8g.large, r8g.xlarge, r8i.2xlarge, r8i.large, r8i.xlarge The cluster's name, as customers see it. Also works on a cluster on the company's BYOC data plane. The bootstrap address and Kubernetes objects keep the name it was created with.
3 - 100How Artie grows the brokers' disks before they fill. Takes effect on the next autoscale run; nothing is applied to the cluster.
New disk per broker in GiB: 50 or 80, or any size from 100 to 1024 (1 TiB). A larger size grows the disks online; the size the brokers already have is left as it is. AWS limits how often a disk can change, so storage grows at most 3 times in 24 hours (429 otherwise).
50 <= x <= 1024Whether every topic is tiered, days on disk, and disk cap per partition. New days on disk or a new cap applies to the running brokers without a restart.
Response
OK
Whether this is a Managed Kafka cluster that Artie runs for the company.
draft, active, deploying, failed, deleted Whether Artie turns tiered storage on for every topic of a Managed Kafka cluster that doesn't choose otherwise.
Whether tiered storage is available on this cluster. Artie turns it on once the cluster's brokers have tiered storage.
When the cluster row last changed. A deploying cluster can be retried 20 minutes after this.
When a deleted Managed Kafka cluster's data is deleted for good. It can be restored until then. Only set while status is deleted.
The step a Managed Kafka cluster's deploy is on. Only set while status is deploying, or failed, where it is the step the deploy stopped at.
applying_custom_resources, issuing_tls_certificate, provisioning_broker_nodes, starting_kafka_brokers, creating_load_balancer Disk cap per partition: the most broker disk each partition of the cluster's tiered topics keeps, in bytes (log.local.retention.bytes). Missing when there is no cap.
Days on disk: how long the cluster's tiered topics keep data on broker disks, in ms (log.local.retention.ms). Missing on clusters made before it, whose brokers use Kafka's default.
A Managed Kafka cluster's configuration. Only set when isManaged, and only by Get a Kafka cluster.
Whether a deleted Managed Kafka cluster is still being taken down. It can't be restored until that finishes. Only set in the clusters list.
Why the last deploy of a Managed Kafka cluster failed. Only set while status is failed.
A Managed Kafka cluster's storage autoscaling. Only set when isManaged.
When the sync last made a Managed Kafka cluster's topics match the tiered storage choice for each. Missing until its first run.

