[
https://issues.apache.org/jira/browse/IGNITE-20148?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexander Lapin updated IGNITE-20148:
-------------------------------------
Description:
h3. Motivation
Currently, locks are released on primary when cleanup is replicated over
majority
{code:java}
private CompletableFuture<Void> processTxCleanupAction(TxCleanupReplicaRequest
request) {
...
return raftClient
.run(txCleanupCmd)
.thenCompose(ignored ->
allOffFuturesExceptionIgnored(txReadFutures, request)
.thenRun(() -> releaseTxLocks(request.txId())));
...
} {code}
That is actually incorrect, because it's possible that primary won't be a part
of majority, meaning that we will release lock still having writeIntent
locally. Generally speaking that should be resovled by implementing writeIntent
resolution for RW transactions
> Explicit writeIntent cleanup on primary replica
> -----------------------------------------------
>
> Key: IGNITE-20148
> URL: https://issues.apache.org/jira/browse/IGNITE-20148
> Project: Ignite
> Issue Type: Bug
> Reporter: Alexander Lapin
> Priority: Major
> Time Spent: 10m
> Remaining Estimate: 0h
>
> h3. Motivation
> Currently, locks are released on primary when cleanup is replicated over
> majority
> {code:java}
> private CompletableFuture<Void>
> processTxCleanupAction(TxCleanupReplicaRequest request) {
> ...
> return raftClient
> .run(txCleanupCmd)
> .thenCompose(ignored ->
> allOffFuturesExceptionIgnored(txReadFutures, request)
> .thenRun(() -> releaseTxLocks(request.txId())));
> ...
> } {code}
> That is actually incorrect, because it's possible that primary won't be a
> part of majority, meaning that we will release lock still having writeIntent
> locally. Generally speaking that should be resovled by implementing
> writeIntent resolution for RW transactions
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
