org.apache.hadoop.hbase.master.procedure

Class ServerCrashProcedure

java.lang.Object

org.apache.hadoop.hbase.procedure2.Procedure<TEnvironment>

org.apache.hadoop.hbase.procedure2.StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

org.apache.hadoop.hbase.master.procedure.ServerCrashProcedure

All Implemented Interfaces:

Comparable<Procedure<MasterProcedureEnv>>, ServerProcedureInterface

Direct Known Subclasses:

HBCKServerCrashProcedure

@InterfaceAudience.Private
public class ServerCrashProcedure
extends StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>
implements ServerProcedureInterface

Handle crashed server. This is a port to ProcedureV2 of what used to be euphemistically called ServerShutdownHandler.

The procedure flow varies dependent on whether meta is assigned and if we are to split logs.

We come in here after ServerManager has noticed a server has expired. Procedures queued on the rpc should have been notified about fail and should be concurrently getting themselves ready to assign elsewhere.

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.hadoop.hbase.procedure2.StateMachineProcedure

StateMachineProcedure.Flow

Nested classes/interfaces inherited from class org.apache.hadoop.hbase.procedure2.Procedure

Procedure.LockState

Nested classes/interfaces inherited from interface org.apache.hadoop.hbase.master.procedure.ServerProcedureInterface

ServerProcedureInterface.ServerOperationType

Field Summary

Fields

Modifier and Type	Field and Description
private boolean	carryingMeta
private org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState	currentRunningState
static boolean	DEFAULT_MASTER_SCP_RETAIN_ASSIGNMENT Default value of MASTER_SCP_RETAIN_ASSIGNMENT
private static org.slf4j.Logger	LOG
static String	MASTER_SCP_RETAIN_ASSIGNMENT Configuration parameter to enable/disable the retain region assignment during ServerCrashProcedure.
private boolean	notifiedDeadServer Whether DeadServer knows that we are processing it.
private List<RegionInfo>	regionsOnCrashedServer Regions that were on the crashed server.
private ServerName	serverName Name of the crashed server to process.
private boolean	shouldSplitWal
private MonitoredTask	status

Fields inherited from class org.apache.hadoop.hbase.procedure2.StateMachineProcedure

stateCount

Fields inherited from class org.apache.hadoop.hbase.procedure2.Procedure

NO_PROC_ID, NO_TIMEOUT

Constructor Summary

Constructors

Constructor and Description
ServerCrashProcedure() Used when deserializing from a procedure store; we'll construct one of these then call #deserializeStateData(InputStream).
ServerCrashProcedure(MasterProcedureEnv env, ServerName serverName, boolean shouldSplitWal, boolean carryingMeta) Call this constructor queuing up a Procedure.

Method Summary

Modifier and Type	Method and Description
protected boolean	abort(MasterProcedureEnv env) The abort() call is asynchronous and each procedure must decide how to deal with it, if they want to be abortable.
protected Procedure.LockState	acquireLock(MasterProcedureEnv env) The user should override this method if they need a lock on an Entity.
private void	assignRegions(MasterProcedureEnv env, List<RegionInfo> regions) Assign the regions on the crashed RS to other Rses.
private void	cleanupSplitDir(MasterProcedureEnv env)
private Procedure[]	createSplittingWalProcedures(MasterProcedureEnv env, boolean splitMeta)
protected void	deserializeStateData(ProcedureStateSerializer serializer) Called on store load to allow the user to decode the previously serialized state.
protected StateMachineProcedure.Flow	executeFromState(MasterProcedureEnv env, org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state) called to perform a single step of the specified 'state' of the procedure
private boolean	filterDefaultMetaRegions()
protected org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState	getInitialState() Return the initial state object that will be used for the first call to executeFromState().
protected ProcedureMetrics	getProcedureMetrics(MasterProcedureEnv env) Override this method to provide procedure specific counters for submitted count, failed count and time histogram.
String	getProcName()
(package private) List<RegionInfo>	getRegionsOnCrashedServer(MasterProcedureEnv env) Returns List of Regions on crashed server.
ServerName	getServerName() Returns Name of this server instance.
ServerProcedureInterface.ServerOperationType	getServerOperationType() Given an operation type we can take decisions about what to do with pending operations.
protected org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState	getState(int stateId) Convert an ordinal (or state id) to an Enum (or more descriptive) state object.
protected int	getStateId(org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state) Convert the Enum (or more descriptive) state object to an ordinal (or state id).
boolean	hasMetaTableRegion() Returns True if this server has an hbase:meta table region.
protected boolean	holdLock(MasterProcedureEnv env) Used to keep the procedure lock even when the procedure is yielding or suspended.
private boolean	isDefaultMetaRegion(RegionInfo hri)
boolean	isInRecoverMetaState()
protected boolean	isMatchingRegionLocation(RegionStateNode rsn) Moved out here so can be overridden by the HBCK fix-up SCP to be less strict about what it will tolerate as a 'match'.
private boolean	isSplittingDone(MasterProcedureEnv env, boolean splitMeta)
protected void	releaseLock(MasterProcedureEnv env) The user should override this method, and release lock if necessary.
protected void	rollbackState(MasterProcedureEnv env, org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state) called to perform the rollback of the specified state
protected void	serializeStateData(ProcedureStateSerializer serializer) The user-level code of the procedure may have some state to persist (e.g.
protected boolean	shouldWaitClientAck(MasterProcedureEnv env) By default, the executor will keep the procedure result around util the eviction TTL is expired.
void	toStringClassDetails(StringBuilder sb) Extend the toString() information with the procedure details e.g.
(package private) void	updateProgress(boolean updateState)
static void	updateProgress(MasterProcedureEnv env, long parentId)
private void	zkCoordinatedSplitLogs(MasterProcedureEnv env) Split logs using 'classic' zk-based coordination.
private void	zkCoordinatedSplitMetaLogs(MasterProcedureEnv env) Split hbase:meta logs using 'classic' zk-based coordination.

Methods inherited from class org.apache.hadoop.hbase.procedure2.StateMachineProcedure

addChildProcedure, execute, failIfAborted, getCurrentState, getCurrentStateId, getCycles, isEofState, isRollbackSupported, isYieldAfterExecutionStep, isYieldBeforeExecuteFromState, rollback, setNextState, toStringState

Methods inherited from class org.apache.hadoop.hbase.procedure2.Procedure

addStackIndex, afterReplay, beforeReplay, bypass, compareTo, completionCleanup, doExecute, doRollback, elapsedTime, getChildrenLatch, getException, getLastUpdate, getNonceKey, getOwner, getParentProcId, getProcId, getProcIdHashCode, getResult, getRootProcedureId, getRootProcId, getStackIndexes, getState, getSubmittedTime, getTimeout, getTimeoutTimestamp, hasChildren, hasException, hasLock, hasOwner, hasParent, hasTimeout, haveSameParent, incChildrenLatch, isBypass, isFailed, isFinished, isInitializing, isLockedWhenLoading, isRunnable, isSuccess, isWaiting, removeStackIndex, setAbortFailure, setChildrenLatch, setFailure, setFailure, setLastUpdate, setNonceKey, setOwner, setOwner, setParentProcId, setProcId, setResult, setRootProcId, setStackIndexes, setState, setSubmittedTime, setTimeout, setTimeoutFailure, skipPersistence, toString, toStringClass, toStringDetails, toStringSimpleSB, updateMetricsOnFinish, updateMetricsOnSubmit, updateTimestamp, waitInitialized, wasExecuted

Methods inherited from class java.lang.Object

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

Field Detail

LOG

private static final org.slf4j.Logger LOG

MASTER_SCP_RETAIN_ASSIGNMENT

public static final String MASTER_SCP_RETAIN_ASSIGNMENT

Configuration parameter to enable/disable the retain region assignment during ServerCrashProcedure.

By default retain assignment is disabled which makes the failover faster and improve the availability; useful for cloud scenario where region block locality is not important. Enable this when RegionServers are deployed on same host where Datanode are running, this will improve read performance due to local read.

see HBASE-24900 for more details.

See Also:

Constant Field Values

DEFAULT_MASTER_SCP_RETAIN_ASSIGNMENT

public static final boolean DEFAULT_MASTER_SCP_RETAIN_ASSIGNMENT

Default value of MASTER_SCP_RETAIN_ASSIGNMENT

See Also:

Constant Field Values

serverName

private ServerName serverName

Name of the crashed server to process.

notifiedDeadServer

private boolean notifiedDeadServer

Whether DeadServer knows that we are processing it.

regionsOnCrashedServer

private List<RegionInfo> regionsOnCrashedServer

Regions that were on the crashed server.

carryingMeta

private boolean carryingMeta

shouldSplitWal

private boolean shouldSplitWal

status

private MonitoredTask status

currentRunningState

private org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState currentRunningState

Constructor Detail

ServerCrashProcedure

public ServerCrashProcedure(MasterProcedureEnv env,
                            ServerName serverName,
                            boolean shouldSplitWal,
                            boolean carryingMeta)

Call this constructor queuing up a Procedure.

Parameters:

serverName - Name of the crashed server.

shouldSplitWal - True if we should split WALs as part of crashed server processing.

carryingMeta - True if carrying hbase:meta table region.

ServerCrashProcedure

public ServerCrashProcedure()

Used when deserializing from a procedure store; we'll construct one of these then call #deserializeStateData(InputStream). Do not use directly.

Method Detail

isInRecoverMetaState

public boolean isInRecoverMetaState()

executeFromState

protected StateMachineProcedure.Flow executeFromState(MasterProcedureEnv env,
                                                      org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state)
                                               throws ProcedureSuspendedException,
                                                      ProcedureYieldException

Description copied from class: StateMachineProcedure

called to perform a single step of the specified 'state' of the procedure

Specified by:

executeFromState in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

state - state to execute

Returns:

Flow.NO_MORE_STATE if the procedure is completed, Flow.HAS_MORE_STATE if there is another step.

Throws:

ProcedureSuspendedException

ProcedureYieldException

getRegionsOnCrashedServer

List<RegionInfo> getRegionsOnCrashedServer(MasterProcedureEnv env)

Returns List of Regions on crashed server.

cleanupSplitDir

private void cleanupSplitDir(MasterProcedureEnv env)

isSplittingDone

private boolean isSplittingDone(MasterProcedureEnv env,
                                boolean splitMeta)

createSplittingWalProcedures

private Procedure[] createSplittingWalProcedures(MasterProcedureEnv env,
                                                 boolean splitMeta)
                                          throws IOException

Throws:

IOException

filterDefaultMetaRegions

private boolean filterDefaultMetaRegions()

isDefaultMetaRegion

private boolean isDefaultMetaRegion(RegionInfo hri)

zkCoordinatedSplitMetaLogs

private void zkCoordinatedSplitMetaLogs(MasterProcedureEnv env)
                                 throws IOException

Split hbase:meta logs using 'classic' zk-based coordination. Superceded by procedure-based WAL splitting.

Throws:

IOException

See Also:

createSplittingWalProcedures(MasterProcedureEnv, boolean)

zkCoordinatedSplitLogs

private void zkCoordinatedSplitLogs(MasterProcedureEnv env)
                             throws IOException

Split logs using 'classic' zk-based coordination. Superceded by procedure-based WAL splitting.

Throws:

IOException

See Also:

createSplittingWalProcedures(MasterProcedureEnv, boolean)

updateProgress

void updateProgress(boolean updateState)

rollbackState

protected void rollbackState(MasterProcedureEnv env,
                             org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state)
                      throws IOException

Description copied from class: StateMachineProcedure

called to perform the rollback of the specified state

Specified by:

rollbackState in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

state - state to rollback

Throws:

IOException - temporary failure, the rollback will retry later

getState

protected org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState getState(int stateId)

Description copied from class: StateMachineProcedure

Convert an ordinal (or state id) to an Enum (or more descriptive) state object.

Specified by:

getState in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

Parameters:

stateId - the ordinal() of the state enum (or state id)

Returns:

the state enum object

getStateId

protected int getStateId(org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState state)

Description copied from class: StateMachineProcedure

Convert the Enum (or more descriptive) state object to an ordinal (or state id).

Specified by:

getStateId in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

Parameters:

state - the state enum object

Returns:

stateId the ordinal() of the state enum (or state id)

getInitialState

protected org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState getInitialState()

Description copied from class: StateMachineProcedure

Return the initial state object that will be used for the first call to executeFromState().

Specified by:

getInitialState in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

Returns:

the initial state enum object

abort

protected boolean abort(MasterProcedureEnv env)

Description copied from class: Procedure

The abort() call is asynchronous and each procedure must decide how to deal with it, if they want to be abortable. The simplest implementation is to have an AtomicBoolean set in the abort() method and then the execute() will check if the abort flag is set or not. abort() may be called multiple times from the client, so the implementation must be idempotent.

NOTE: abort() is not like Thread.interrupt(). It is just a notification that allows the procedure implementor abort.

Overrides:

abort in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

acquireLock

protected Procedure.LockState acquireLock(MasterProcedureEnv env)

Description copied from class: Procedure

The user should override this method if they need a lock on an Entity. A lock can be anything, and it is up to the implementor. The Procedure Framework will call this method just before it invokes Procedure.execute(Object). It calls Procedure.releaseLock(Object) after the call to execute.

If you need to hold the lock for the life of the Procedure -- i.e. you do not want any other Procedure interfering while this Procedure is running, see Procedure.holdLock(Object).

Example: in our Master we can execute request in parallel for different tables. We can create t1 and create t2 and these creates can be executed at the same time. Anything else on t1/t2 is queued waiting that specific table create to happen.

There are 3 LockState:

• LOCK_ACQUIRED should be returned when the proc has the lock and the proc is ready to execute.
• LOCK_YIELD_WAIT should be returned when the proc has not the lock and the framework should take care of readding the procedure back to the runnable set for retry
• LOCK_EVENT_WAIT should be returned when the proc has not the lock and someone will take care of readding the procedure back to the runnable set when the lock is available.

Overrides:

acquireLock in class Procedure<MasterProcedureEnv>

Returns:

the lock state as described above.

releaseLock

protected void releaseLock(MasterProcedureEnv env)

Description copied from class: Procedure

The user should override this method, and release lock if necessary.

Overrides:

releaseLock in class Procedure<MasterProcedureEnv>

toStringClassDetails

public void toStringClassDetails(StringBuilder sb)

Description copied from class: Procedure

Extend the toString() information with the procedure details e.g. className and parameters

Overrides:

toStringClassDetails in class Procedure<MasterProcedureEnv>

Parameters:

sb - the string builder to use to append the proc specific information

getProcName

public String getProcName()

Overrides:

getProcName in class Procedure<MasterProcedureEnv>

serializeStateData

protected void serializeStateData(ProcedureStateSerializer serializer)
                           throws IOException

Description copied from class: Procedure

The user-level code of the procedure may have some state to persist (e.g. input arguments or current position in the processing state) to be able to resume on failure.

Overrides:

serializeStateData in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

Parameters:

serializer - stores the serializable state

Throws:

IOException

deserializeStateData

protected void deserializeStateData(ProcedureStateSerializer serializer)
                             throws IOException

Description copied from class: Procedure

Called on store load to allow the user to decode the previously serialized state.

Overrides:

deserializeStateData in class StateMachineProcedure<MasterProcedureEnv,org.apache.hadoop.hbase.shaded.protobuf.generated.MasterProcedureProtos.ServerCrashState>

Parameters:

serializer - contains the serialized state

Throws:

IOException

getServerName

public ServerName getServerName()

Description copied from interface: ServerProcedureInterface

Returns Name of this server instance.

Specified by:

getServerName in interface ServerProcedureInterface

hasMetaTableRegion

public boolean hasMetaTableRegion()

Description copied from interface: ServerProcedureInterface

Returns True if this server has an hbase:meta table region.

Specified by:

hasMetaTableRegion in interface ServerProcedureInterface

getServerOperationType

public ServerProcedureInterface.ServerOperationType getServerOperationType()

Description copied from interface: ServerProcedureInterface

Given an operation type we can take decisions about what to do with pending operations. e.g. if we get a crash handler and we have some assignment operation pending we can abort those operations.

Specified by:

getServerOperationType in interface ServerProcedureInterface

Returns:

the operation type that the procedure is executing.

shouldWaitClientAck

protected boolean shouldWaitClientAck(MasterProcedureEnv env)

Description copied from class: Procedure

By default, the executor will keep the procedure result around util the eviction TTL is expired. The client can cut down the waiting time by requesting that the result is removed from the executor. In case of system started procedure, we can force the executor to auto-ack.

Overrides:

shouldWaitClientAck in class Procedure<MasterProcedureEnv>

Parameters:

env - the environment passed to the ProcedureExecutor

Returns:

true if the executor should wait the client ack for the result. Defaults to return true.

isMatchingRegionLocation

protected boolean isMatchingRegionLocation(RegionStateNode rsn)

Moved out here so can be overridden by the HBCK fix-up SCP to be less strict about what it will tolerate as a 'match'.

Returns:

True if the region location in rsn matches that of this crashed server.

assignRegions

private void assignRegions(MasterProcedureEnv env,
                           List<RegionInfo> regions)
                    throws IOException

Assign the regions on the crashed RS to other Rses.

In this method we will go through all the RegionStateNodes of the give regions to find out whether there is already an TRSP for the region, if so we interrupt it and let it retry on other server, otherwise we will schedule a TRSP to bring the region online.

We will also check whether the table for a region is enabled, if not, we will skip assigning it.

Throws:

IOException

getProcedureMetrics

protected ProcedureMetrics getProcedureMetrics(MasterProcedureEnv env)

Description copied from class: Procedure

Override this method to provide procedure specific counters for submitted count, failed count and time histogram.

Overrides:

getProcedureMetrics in class Procedure<MasterProcedureEnv>

Parameters:

env - The environment passed to the procedure executor

Returns:

Container object for procedure related metric

holdLock

protected boolean holdLock(MasterProcedureEnv env)

Description copied from class: Procedure

Used to keep the procedure lock even when the procedure is yielding or suspended.

Overrides:

holdLock in class Procedure<MasterProcedureEnv>

Returns:

true if the procedure should hold on the lock until completionCleanup()

updateProgress

public static void updateProgress(MasterProcedureEnv env,
                                  long parentId)