ChenSammi commented on code in PR #3403:
URL: https://github.com/apache/ozone/pull/3403#discussion_r870894580
##########
hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/ha/SCMStateMachine.java:
##########
@@ -133,6 +133,15 @@ public CompletableFuture<Message> applyTransaction(
final TransactionContext trx) {
final CompletableFuture<Message> applyTransactionFuture =
new CompletableFuture<>();
+ if (getLifeCycleState().isPausingOrPaused()) {
+ // Statemachine is installation snapshot. Discard any request.
Review Comment:
Ratis will call StateMachine#pause before snapshot install, and then call
StateMachine#reinitialize after statemachine reloaded. In SCM
StateMachine#reinitialize, the statemachine will turn itself into RUNNING
state. All this happens concurrently with other Ratis appendEntities request
handling. So we need to check the statemachine state in applyTransaction.
Refer to the crash file attached.
In OM statemachine implementation, we don't check this statemachine state in
applyTransaction. Because all transaction output data go into OMDoubleBuffer,
and we stop OMDoubleBuffer before we reload DB, and restart OMDoubleBuffer
after that.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
