imbajin commented on code in PR #2871:
URL: https://github.com/apache/hugegraph/pull/2871#discussion_r4098105266
##########
hugegraph-store/hg-store-core/src/main/java/org/apache/hugegraph/store/raft/PartitionStateMachine.java:
##########
@@ -105,6 +105,9 @@ public void onApply(Iterator iter) {
done.getOperation().getOp(),
done.getOperation().getReq());
}
+ iter.setErrorAndRollback(1, new
Status(RaftError.ESTATEMACHINE, t.getMessage()));
Review Comment:
⚠️ This catch also covers `done.run(Status.OK())` in the leader path. The
`Change_Shard` handler updates its task and starts follow-up work before
invoking the response closure, so a thrown callback is treated as an apply
failure after side effects have started and can fail-stop this FSM. Please keep
callback exceptions out of the rollback path.
##########
hugegraph-store/hg-store-core/src/main/java/org/apache/hugegraph/store/raft/PartitionStateMachine.java:
##########
@@ -105,6 +105,9 @@ public void onApply(Iterator iter) {
done.getOperation().getOp(),
done.getOperation().getReq());
}
+ iter.setErrorAndRollback(1, new
Status(RaftError.ESTATEMACHINE, t.getMessage()));
Review Comment:
⚠️ This records a terminal state-machine error for the committed entry, but
`PartitionEngine.onError()` immediately restarts the engine; `shutdown()` keeps
the same log/meta directories that `init()` reopens. A deterministic failure
will replay the entry and restart again, leaving the partition unavailable.
Please stop automatic restart for this error or add bounded recovery that does
not replay the poison entry.
--
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]
