IGNITE-23748 Lock LWM when executing RO operation on data node #4974
Conversation
rpuch
force-pushed
the
ignite-23748
branch
3 times, most recently
from
056d370 to
9ff9923
Compare
* When executing an operation in an RO transaction (explicit or implicit), attempt to lock LWM on the data node where it's executed * If lock attempt fails, throw an exception with a specific error code * When cleaning up after an RO transaction had been finished, unlock LWM on each node where such a cleanup happens * For direct RO operations (which happen in implicit RO transactions), unlock LWM right after the read has been done on the data node
modules/client/src/test/java/org/apache/ignite/client/fakes/FakeInternalTable.java
Show resolved
Hide resolved
...-engine/src/integrationTest/java/org/apache/ignite/internal/sql/sqllogic/ItSqlLogicTest.java
Show resolved
Hide resolved
...n/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java
Show resolved
Hide resolved
| @@ -534,7 +540,7 @@ private CompletableFuture<?> processRequest(ReplicaRequest request, @Nullable Bo | |||
| // Don't need to validate schema. | |||
| if (opTs == null) { | |||
| assert opTsIfDirectRo == null; | |||
| return processOperationRequestWithTxRwCounter(senderId, request, isPrimary, null, leaseStartTime); | |||
| return processOperationRequestWithWrappingLogic(senderId, request, isPrimary, null, leaseStartTime); | |||
There was a problem hiding this comment.
I don't like the postfix "WithWrappingLogic", wrapping is very general term and thus such postfix says nothing.
There was a problem hiding this comment.
This method now adds wrapping related to:
- RW operation counting
- RO operation LWM locking/unlocking
The old name only mentioned 1 and now became 'not completely true'. Adding another suffix to mention RO+LWM to the existing one would produce a monstrosity. That's why I chose a generic name that says that it adds some wrapping.
Do you have a specific idea on how to improve this?
There was a problem hiding this comment.
Do you have a specific idea on how to improve this?
Nothing good, e.g. processOperationRequestWithTxResourceManagementWrappingLogic. Up to you.
There was a problem hiding this comment.
Renamed to processOperationRequestWithTxOperationManagementLogic() (one word shorter than you suggested as it was too ugly wrt line breaks)
...n/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java
Show resolved
Hide resolved
...n/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java
Show resolved
Hide resolved
...n/java/org/apache/ignite/internal/table/distributed/replicator/PartitionReplicaListener.java
Show resolved
Hide resolved
...grationTest/java/org/apache/ignite/internal/tx/readonly/ItReadOnlyTxAndLowWatermarkTest.java
Show resolved
Hide resolved
...grationTest/java/org/apache/ignite/internal/tx/readonly/ItReadOnlyTxAndLowWatermarkTest.java
Show resolved
Hide resolved
...les/transactions/src/main/java/org/apache/ignite/internal/tx/impl/ResourceVacuumManager.java
Outdated
Show resolved
Hide resolved
| @Nullable | ||
| UUID transactionId(); | ||
|
|
||
| /** Inconsistent ID of transaction to which this operation belongs. */ |
There was a problem hiding this comment.
Inconsistent ID of transaction coordinator I guess.
https://issues.apache.org/jira/browse/IGNITE-23748