org.corfudb.runtime.view

Class ManagementView

java.lang.Object

org.corfudb.runtime.view.AbstractView

org.corfudb.runtime.view.ManagementView

public class ManagementView
extends AbstractView

A view of the Management Service to manage reconfigurations of the Corfu Cluster.

Created by zlokhandwala on 11/20/17.

Nested Class Summary

Nested classes/interfaces inherited from class org.corfudb.runtime.view.AbstractView

AbstractView.LayoutFunction<V,R,A extends Throwable,B extends Throwable,C extends Throwable,D extends Throwable>

Constructor Summary

Constructors

Constructor and Description
ManagementView(CorfuRuntime runtime)

Method Summary

Modifier and Type	Method and Description
void	addNode(String endpointToAdd, int retry, Duration timeout, Duration pollPeriod) Add a new node to the existing cluster.
void	forceRemoveNode(String endpointToRemove, int retry, Duration timeout, Duration pollPeriod) Force remove a node from the cluster.
ClusterStatusReport	getClusterStatus() Get the Cluster Status.
void	healNode(String endpointToHeal, int retry, Duration timeout, Duration pollPeriod) Heal an unresponsive node.
void	mergeSegments(String endpointToRestoreRedundancy, int retry, Duration timeout, Duration pollPeriod) Restore redundancy and merge all split segments.
void	removeNode(String endpointToRemove, int retry, Duration timeout, Duration pollPeriod) Remove a node from the cluster.

Methods inherited from class org.corfudb.runtime.view.AbstractView

getCurrentLayout, layoutHelper, layoutHelper

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ManagementView

public ManagementView(@NonNull
                      CorfuRuntime runtime)

Method Detail

removeNode

public void removeNode(@Nonnull
                       String endpointToRemove,
                       int retry,
                       @Nonnull
                       Duration timeout,
                       @Nonnull
                       Duration pollPeriod)

Remove a node from the cluster.

Parameters:

endpointToRemove - Endpoint of the node to be removed from the cluster.

retry - the number of times to retry a workflow if it fails

timeout - total time to wait before the workflow times out

pollPeriod - the poll interval to check whether a workflow completed or not

Throws:

WorkflowResultUnknownException - when the side affect of the operation can't be determined

WorkflowException - when the remove operation fails

forceRemoveNode

public void forceRemoveNode(@Nonnull
                            String endpointToRemove,
                            int retry,
                            @Nonnull
                            Duration timeout,
                            @Nonnull
                            Duration pollPeriod)

Force remove a node from the cluster.

Parameters:

endpointToRemove - Endpoint of the node to be removed from the cluster.

retry - the number of times to retry a workflow if it fails

timeout - total time to wait before the workflow times out

pollPeriod - the poll interval to check whether a workflow completed or not

Throws:

WorkflowResultUnknownException - when the side affect of the operation can't be determined

WorkflowException - when the remove operation fails

addNode

public void addNode(@Nonnull
                    String endpointToAdd,
                    int retry,
                    @Nonnull
                    Duration timeout,
                    @Nonnull
                    Duration pollPeriod)

Add a new node to the existing cluster.

Parameters:

endpointToAdd - Endpoint of the new node to be added to the cluster.

retry - the number of times to retry a workflow if it fails

timeout - total time to wait before the workflow times out

pollPeriod - the poll interval to check whether a workflow completed or not

Throws:

WorkflowResultUnknownException - when the side affect of the operation can't be determined

healNode

public void healNode(@Nonnull
                     String endpointToHeal,
                     int retry,
                     @Nonnull
                     Duration timeout,
                     @Nonnull
                     Duration pollPeriod)

Heal an unresponsive node.

Parameters:

endpointToHeal - Endpoint of the new node to be healed in the cluster.

retry - the number of times to retry a workflow if it fails

timeout - total time to wait before the workflow times out

pollPeriod - the poll interval to check whether a workflow completed or not

Throws:

WorkflowResultUnknownException - when the side affect of the operation can't be determined

mergeSegments

public void mergeSegments(@Nonnull
                          String endpointToRestoreRedundancy,
                          int retry,
                          @Nonnull
                          Duration timeout,
                          @Nonnull
                          Duration pollPeriod)

Restore redundancy and merge all split segments.

Parameters:

endpointToRestoreRedundancy - Endpoint whose redundancy is to be restored.

retry - the number of times to retry a workflow if it fails

timeout - total time to wait before the workflow times out

pollPeriod - the poll interval to check whether a workflow completed or not

Throws:

WorkflowResultUnknownException - when the side affect of the operation can't be determined

getClusterStatus

public ClusterStatusReport getClusterStatus()

Get the Cluster Status. This is reported as follows: - (1) The status of the cluster itself (regardless of clients connectivity) as reflected in the layout. This information is presented along each node's status (up, down, db_sync). It is important to note that as the cluster state is obtained from the layout, when quorum is not available (majority of nodes) there are lower guarantees on the reliability of this state. For example, in the absence of quorum the system might be in an unstable state which cannot converge due to lack of consensus. This is reflected in the cluster status report as clusterStatusReliability. - (2) The connectivity status of the client to every node in the cluster, i.e., can the client connect to the cluster. This will be obtained by ping(ing) every node and show as RESPONSIVE, for successful connections or UNRESPONSIVE for clients unable to communicate. In this sense a cluster can be STABLE with all nodes UP, while not being available for a client, as all connections from the client to the cluster nodes are down, showing in this case connectivity status to all nodes as UNRESPONSIVE. The ClusterStatusReport consists of the following: CLUSTER-SPECIFIC STATUS - clusterStatus: the cluster status a perceived by the system's layout. STABLE, DEGRADED, DB_SYNCING or UNAVAILABLE - nodeStatusMap: each node's status as perceived by the system's layout. (UP, DOWN or DB_SYNC) - Cluster Status Reliability: STRONG_QUORUM, WEAK_NO_QUORUM or UNAVAILABLE CLIENT-CLUSTER SPECIFIC STATUS: - clientServerConnectivityStatusMap: the connectivity status of this client to the cluster. (RESPONSIVE, UNRESPONSIVE).

Returns:

ClusterStatusReport