The Provisioning Life Cycle
Objects (i.e. nodes, DC, Cluster) move through the following states under normal operations:
GENESIS ⇨ PROVISIONING ⇨ PROVISIONED ⇨ JOINING (optional) ⇨ RUNNING ⇨ DELETING ⇨ DELETED
It is important to think of a cluster Object as a “tree-like” structure. With the Cluster object at the top, under that you can have one or more DCs, with each DC (object) having three or more nodes. Now, because the cluster Object is tree-like, the running state propagates upwards. For example a DC will be in a PROVISIONED state until all the nodes under it reach a RUNNING state, then its state will be RUNNING.
If you are using the API the cluster status call allows you to programmatically determine the status of the objects in a cluster.
In both DEFERRED and FAILED cases we will notify you if any action is required on your end, and look to resolve the cause as soon as possible.
|Provisioning Status||Description||Customer Action|
|GENESIS||Before the provisioning operation kicks off, everything is in GENESIS state until the pre-provision checks are completed (provisioning, billing, etc.). As soon as an object is picked up for creation, it moves to PROVISIONING state.||There is no action to take for GENESIS.|
|PROVISIONING||When the nodes are created with the cloud service provider, they are in PROVISIONING state until the node has been fully configured, the required applications are installed, and the primary bundle is ready to be started.||There is no action to take for PROVISIONING under normal operations. |
As a conservative guide for the PROVISIONING state, allow approximately 5 min per node in the cluster. If this time is exceeded then contact Instaclustr support for assistance.
|PROVISIONED||All the components of an object provisioning have completed, and the primary bundle has been started, but has not yet reached a RUNNING state. Some applications, such as Cassandra, should generally only have 1 node at a time in the PROVISIONED state.||There is no action to take for PROVISIONED under normal operations.|
As a conservative guide for the PROVISIONED state, allow approximately 5 min per node in the cluster to progress to RUNNING state.
|RUNNING||Once we receive a heartbeat from the node telling our central monitoring server that everything is up and running, the node state moves to RUNNING. Once an object reaches RUNNING state it stays in that state until it is deleted (i.e. this is not a live service status).||There is no action to take for RUNNING. |
If a cluster is in this state and services do not appear to be available then contact Instaclustr support.
|DELETING||Once an object is requested to be deleted, it moves to the DELETING state. Once marked for deletion, the object is removed from view.||There is no action to take for DELETING. |
Instaclustr support will be alerted if deletion fails and advise if any actions are necessary. Objects will normally be in DELETING state for less that 15 minutes (very large clusters may take longer).
|DELETED||The state once an object has completed being DELETING.||There is no action to take for DELETED.|
|JOINING||When a node is joining the cluster. In this state it is streaming data from other nodes in the cluster. |
Clusters with small amounts of data will have nodes join quite quickly.
|There is no action to take for JOINING. A conservative estimate for JOINING Cassandra nodes is to allow 1 day per 1 TB of data. If customer action is required for large nodes, support will be in contact.|
|DEFERRED||DEFERRED is a special state, it indicates that an issue has occurred that requires intervention and when resolved, our provisioner can automatically pick up where it left off. Examples of this could include failing pre-provision checks or requests that required our support team input, such as adding a node to a cluster.||If customer action is required to resume provisioning, support will be in contact.|
|FAILED||FAILED status indicates an unexpected error occurred at some point in the provisioning process, and action is required to resolve it. Most of the time this is relating to an issue occurring with the cloud provider, such as a misconfiguration for external accounts or limits being reached||If customer action is required, support will be in contact.|