com.alipay.sofa.jraft

接口 Iterator

所有超级接口:

Iterator<ByteBuffer>

public interface Iterator
extends Iterator<ByteBuffer>

Iterator over a batch of committed tasks.

作者:

boyan (boyan@alibaba-inc.com) 2018-Apr-03 3:20:15 PM

另请参阅:

StateMachine.onApply(Iterator)

方法概要

限定符和类型	方法和说明
Closure	done() If done() is non-NULL, you must call done()->Run() after applying this task no matter this operation succeeds or fails, otherwise the corresponding resources would leak.
ByteBuffer	getData() Return the data whose content is the same as what was passed to Node#apply(Task) in the leader node.
long	getIndex() Return a unique and monotonically increasing identifier of the current task: - Uniqueness guarantees that committed tasks in different peers with the same index are always the same and kept unchanged
long	getTerm() Returns the term of the leader which to task was applied to.
void	setErrorAndRollback(long ntail, Status st) Invoked when some critical error occurred.

从接口继承的方法 java.util.Iterator

forEachRemaining, hasNext, next, remove

方法详细资料

getData

ByteBuffer getData()

Return the data whose content is the same as what was passed to Node#apply(Task) in the leader node.

getIndex

long getIndex()

Return a unique and monotonically increasing identifier of the current task: - Uniqueness guarantees that committed tasks in different peers with the same index are always the same and kept unchanged. - Monotonicity guarantees that for any index pair i, j (i < j), task at index |i| must be applied before task at index |j| in all the peers from the group.

getTerm

long getTerm()

Returns the term of the leader which to task was applied to.

done

Closure done()

If done() is non-NULL, you must call done()->Run() after applying this task no matter this operation succeeds or fails, otherwise the corresponding resources would leak. If this task is proposed by this Node when it was the leader of this group and the leadership has not changed before this point, done() is exactly what was passed to Node#apply(Task) which may stand for some continuation (such as respond to the client) after updating the StateMachine with the given task. Otherwise done() must be NULL.

setErrorAndRollback

void setErrorAndRollback(long ntail,
                         Status st)

Invoked when some critical error occurred. And we will consider the last |ntail| tasks (starting from the last iterated one) as not applied. After this point, no further changes on the StateMachine as well as the Node would be allowed and you should try to repair this replica or just drop it.

参数:

ntail - the number of tasks (starting from the last iterated one) considered as not to be applied.

st - Status to describe the detail of the error.