DNode Cluster Management landing page
|
|
Access this by selecting the Designer Studio > Decisioning > Infrastructure > DNode Cluster Management menu item.
The DNode Cluster Management landing page provides an overview of the DNode cluster infrastructure and allows you to manage the and operations participation of Pega 7 Platform nodes in this infrastructure. Pega 7 Platform nodes that are part of this infrastructure are considered DNodes (Data Nodes).
Available in applications that have access to the Decision Management rule sets or in applications built on PegaDM.
About DNodes
A DNode consists of a standard Pega 7 Platform installation augmented by an embedded Cassandra database process. DNodes rely on the Hazelcast Pega 7 Platform clustering implementation and are capable of storing high volumes of data at high speed. By creating a cluster of DNodes, you can leverage Cassandra's attributes to protect data from failure, as well as provide horizontal scalability. As a Pega 7 Platform installation, the DNode is capable to make specialized decisions in real time, allowing you to take data from different sources that are continuously available (or not), and making it available for real time and batch processing. The embedded nature of the persistence solution means that it is also optimized for batch processing.
DNodes Management
The DNode Management overview provides information about the nodes in the DNode cluster, and the possible actions on a DNode.
- Status: displays the status of nodes in the DNode cluster. The status of a DNode is ONLINE-<state>, in which case the Pega 7 Platform server is online, or OFFLINE-<state>.
The available statuses| Status | Description |
|---|
| ONLINE-PENDING-JOINING | The request was made for the node to join the DNode cluster. |
| ONLINE-JOINING | Node is in the process of being enabled, and the server is joining the DNode cluster. |
| ONLINE-JOINING-FAILED | Node failed to join the DNode cluster. You can try to add the node again. |
| ONLINE-NORMAL | DNode is in normal, functional state. |
| ONLINE-LEAVING | Node is in the process of being removed, and the server is leaving the DNode cluster. |
| ONLINE-MOVING | DNode is in the process of being moved to a new position in the DNode cluster ring. |
| ONLINE-CORRUPTED | The DNode's file system is corrupted. Use the repair operation to make this DNode operational again. |
| ONLINE-REPAIRING | The repair operation is running on the DNode. |
| ONLINE-COMPACTING | The compact operation is running on the DNode. |
| ONLINE-CLEARING | The clean up operation is running on the DNode. |
| ONLINE-DECOMMISSIONED | Server has been decommissioned, and this node is no longer a DNode. |
| ONLINE-UNKNOWN | The status of the DNode is currently unknown. |
| OFFLINE-UNKNOWN | Although the Pega 7 Platform server is offline, the node is still recognized as a DNode. |
- IP Address: displays the node's IP address.
- Node Name: displays the node's name.
- Memory Usage: displays the amount of memory usage, as well as the maximum amount of allocated memory to the JVM.
- Owns: displays the payload taken by this node.
- Node ID: displays the node's ID.
- Action: provides the menu with the possible operations that you can perform on a DNode.
The available actions| Action | Description |
|---|
| Repair | Repairs inconsistencies across all replicas of the data sets in a DNode. Run this operation in the following cases:
- As part of node recovery, when bringing a node back into the cluster after a failure.
- As part of the maintenance of nodes containing data that is not frequently read.
- As part of the update process of nodes that were offline.
|
| Compact | This operation invokes the process by which Cassandra combines multiple physical data files on the file system to improve the performance of partition scans and to reclaim space from deleted data. Compaction is a background operation that runs on a regular basis in the DNode cluster, but you can run it explicitly on one DNode through the compact action. |
| Clean up | Removes keys that no longer belong to the DNode. Keys do not belong to a DNode if data was deleted, or if it was rebalanced within the DNode cluster. |
| Decommission | Remove the node from the DNode cluster. This operation results in transferring the data contained in the DNode storage to other nodes in the DNode cluster. If you want to add a decommissioned node to the DNode cluster, you have to restart the Pega 7 Platform server before adding it again as a DNode. |
Adding and Removing DNodes
The DNode cluster infrastructure detects if there are Pega 7 Platform nodes configured as DNodes. The capability to add DNodes is disabled if the nodes are already part of the DNode cluster, or no nodes in the infrastructure are configured to be DNodes.
Any Pega 7 Platform node that is part of an outer cluster of Pega 7 Platform nodes on the same network based on Halzelcast can be a DNode. Using this landing page, you can manage the membership of Pega 7 Platform nodes in the DNode cluster infrastructure.
- Add node to the DNode infrastructure by selecting the node and clicking Add node. Refresh the landing page. Most likely, the status of the node is still ONLINE-JOINING. The node becomes part of the DNode cluster infrastructure when the status is ONLINE-NORMAL.
- Remove node by locating the appropriate node and, in the Action column, selecting Decommission. Refresh the landing page until you see ONLINE-DECOMISSIONED in the node's status. The node is still online and part of the Hazelcast cluster, but it is no longer a DNode.
Performance Considerations
- During the compaction operation, there is a temporary spike in disk usage and disk I/O. Do not run this operation during peak usage hours on multiple DNodes concurrently as it negatively impacts performance for any process or request accessing the DNode storage.
- The clean up operation has a similar impact on performance as the compaction operation.
- Data is distributed randomly across nodes in the DNode cluster. For this reason, you can expect approximately the same increase in disk space usage at the same rate. If nodes are running out of disk space, your DNode cluster infrastructure needs to be resized, and you should add one more node to the DNode cluster infrastructure.
Related information
Open topic with navigation