[
https://issues.apache.org/jira/browse/HDDS-4059?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Glen Geng updated HDDS-4059:
----------------------------
Description:
applyTransaction should not call trx.getClientRequest(), since client request
will not be replicated from leader to follower.
{code:java}
SCMStateMachine
applyTransaction()
final SCMRatisRequest request = SCMRatisRequest.decode(
trx.getClientRequest().getMessage());{code}
instead, we should call trx.getStateMachineLogEntry()
{code:java}
SCMStateMachine
applyTransaction()
final SCMRatisRequest request = SCMRatisRequest.decode(
Message.valueOf(trx.getStateMachineLogEntry().getLogData()));{code}
since content of client request will be injected to StateMachineEntryProto at
leader,
{code:java}
static StateMachineLogEntryProto toStateMachineLogEntryProto(
RaftClientRequest request, ByteString logData, ByteString stateMachineData)
{
if (logData == null) {
logData = request.getMessage().getContent();
}
return toStateMachineLogEntryProto(request.getClientId(),
request.getCallId(), logData, stateMachineData);
}
{code}
and extracted from log entry at follower.
{code:java}
/**
* Construct a {@link TransactionContext} from a {@link LogEntryProto}.
* Used by followers for applying committed entries to the state machine.
* @param logEntry the log entry to be applied
*/
public TransactionContextImpl(RaftPeerRole serverRole, StateMachine
stateMachine, LogEntryProto logEntry) {
this(serverRole, stateMachine);
this.logEntry = logEntry;
this.smLogEntryProto = logEntry.getStateMachineLogEntry();
}
{code}
was:
applyTransaction should not call trx.getClientRequest(), since client request
will not be replicated from leader to follower.
{code:java}
SCMStateMachine
applyTransaction()
final SCMRatisRequest request = SCMRatisRequest.decode(
trx.getClientRequest().getMessage());{code}
instead, we should call trx.getStateMachineLogEntry()
{code:java}
SCMStateMachine
applyTransaction()
final SCMRatisRequest request = SCMRatisRequest.decode(
Message.valueOf(trx.getStateMachineLogEntry().getLogData()));{code}
since content of client request will be injected to StateMachineEntryProto at
leader.
{code:java}
static StateMachineLogEntryProto toStateMachineLogEntryProto(
RaftClientRequest request, ByteString logData, ByteString stateMachineData)
{
if (logData == null) {
logData = request.getMessage().getContent();
}
return toStateMachineLogEntryProto(request.getClientId(),
request.getCallId(), logData, stateMachineData);
}
{code}
and extracted from log entry at follower.
{code:java}
/**
* Construct a {@link TransactionContext} from a {@link LogEntryProto}.
* Used by followers for applying committed entries to the state machine.
* @param logEntry the log entry to be applied
*/
public TransactionContextImpl(RaftPeerRole serverRole, StateMachine
stateMachine, LogEntryProto logEntry) {
this(serverRole, stateMachine);
this.logEntry = logEntry;
this.smLogEntryProto = logEntry.getStateMachineLogEntry();
}
{code}
> SCMStateMachine::applyTransaction() should not invoke
> TransactionContext.getClientRequest()
> -------------------------------------------------------------------------------------------
>
> Key: HDDS-4059
> URL: https://issues.apache.org/jira/browse/HDDS-4059
> Project: Hadoop Distributed Data Store
> Issue Type: Sub-task
> Components: SCM
> Reporter: Glen Geng
> Assignee: Glen Geng
> Priority: Major
>
> applyTransaction should not call trx.getClientRequest(), since client
> request will not be replicated from leader to follower.
> {code:java}
> SCMStateMachine
> applyTransaction()
> final SCMRatisRequest request = SCMRatisRequest.decode(
> trx.getClientRequest().getMessage());{code}
>
> instead, we should call trx.getStateMachineLogEntry()
> {code:java}
> SCMStateMachine
> applyTransaction()
> final SCMRatisRequest request = SCMRatisRequest.decode(
> Message.valueOf(trx.getStateMachineLogEntry().getLogData()));{code}
>
> since content of client request will be injected to StateMachineEntryProto at
> leader,
> {code:java}
> static StateMachineLogEntryProto toStateMachineLogEntryProto(
> RaftClientRequest request, ByteString logData, ByteString
> stateMachineData) {
> if (logData == null) {
> logData = request.getMessage().getContent();
> }
> return toStateMachineLogEntryProto(request.getClientId(),
> request.getCallId(), logData, stateMachineData);
> }
> {code}
>
> and extracted from log entry at follower.
> {code:java}
> /**
> * Construct a {@link TransactionContext} from a {@link LogEntryProto}.
> * Used by followers for applying committed entries to the state machine.
> * @param logEntry the log entry to be applied
> */
> public TransactionContextImpl(RaftPeerRole serverRole, StateMachine
> stateMachine, LogEntryProto logEntry) {
> this(serverRole, stateMachine);
> this.logEntry = logEntry;
> this.smLogEntryProto = logEntry.getStateMachineLogEntry();
> }
> {code}
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org
