search

neo4j / apoc

Apache License 2.0
88 stars 27 forks source link

Issue with triggers in phase 'after' #97

Open neo-technology-build-agent opened 2 years ago

neo-technology-build-agent commented 2 years ago

Issue by kraney Tuesday Mar 17, 2020 at 23:49 GMT Originally opened as https://github.com/neo4j-contrib/neo4j-apoc-procedures/issues/1450

Expected Behavior (Mandatory)

Adding a trigger in phase 'after' does not disrupt any operations

Actual Behavior (Mandatory)

When I create a trigger in phase 'after' - even when I reduce it to simply doing "RETURN NULL", it causes some transactions to fail with a stack trace like the following:

2020-03-17 23:39:53.368+0000 WARN  Error executing trigger onNewDeviceConfig_PublishForModel in phase after Getting deleted relationship data should have been covered by the tx state
java.lang.IllegalStateException: Getting deleted relationship data should have been covered by the tx state
    at org.neo4j.kernel.impl.coreapi.TxStateTransactionDataSnapshot.relationship(TxStateTransactionDataSnapshot.java:354)
    at org.eclipse.collections.impl.lazy.primitive.CollectLongToObjectIterable$2.next(CollectLongToObjectIterable.java:82)
    at org.neo4j.kernel.impl.util.ValueUtils.asListValue(ValueUtils.java:192)
    at org.neo4j.kernel.impl.util.ValueUtils.of(ValueUtils.java:106)
    at org.neo4j.kernel.impl.util.ValueUtils.asParameterMapValue(ValueUtils.java:249)
    at org.neo4j.kernel.impl.factory.GraphDatabaseFacade.execute(GraphDatabaseFacade.java:404)
    at apoc.trigger.Trigger$TriggerHandler.lambda$executeTriggers$0(Trigger.java:282)
    at java.util.concurrent.ConcurrentHashMap.forEach(ConcurrentHashMap.java:1597)
    at apoc.trigger.Trigger$TriggerHandler.executeTriggers(Trigger.java:272)
    at apoc.trigger.Trigger$TriggerHandler.afterCommit(Trigger.java:305)
    at org.neo4j.kernel.internal.TransactionEventHandlers.afterCommit(TransactionEventHandlers.java:149)
    at org.neo4j.kernel.internal.TransactionEventHandlers.afterCommit(TransactionEventHandlers.java:48)
    at org.neo4j.kernel.impl.api.TransactionHooks.afterCommit(TransactionHooks.java:76)
    at org.neo4j.kernel.impl.api.KernelTransactionImplementation.afterCommit(KernelTransactionImplementation.java:922)
    at org.neo4j.kernel.impl.api.KernelTransactionImplementation.commit(KernelTransactionImplementation.java:745)
    at org.neo4j.kernel.impl.api.KernelTransactionImplementation.closeTransaction(KernelTransactionImplementation.java:594)
    at org.neo4j.internal.kernel.api.Transaction.close(Transaction.java:178)
    at org.neo4j.bolt.v1.runtime.TransactionStateMachine$State.closeTransaction(TransactionStateMachine.java:469)
    at org.neo4j.bolt.v1.runtime.TransactionStateMachine$State$2.commitTransaction(TransactionStateMachine.java:402)
    at org.neo4j.bolt.v1.runtime.TransactionStateMachine.commitTransaction(TransactionStateMachine.java:144)
    at org.neo4j.bolt.v3.runtime.TransactionReadyState.processCommitMessage(TransactionReadyState.java:93)
    at org.neo4j.bolt.v3.runtime.TransactionReadyState.processUnsafe(TransactionReadyState.java:53)
    at org.neo4j.bolt.v3.runtime.FailSafeBoltStateMachineState.process(FailSafeBoltStateMachineState.java:48)
    at org.neo4j.bolt.v1.runtime.BoltStateMachineV1.nextState(BoltStateMachineV1.java:144)
    at org.neo4j.bolt.v1.runtime.BoltStateMachineV1.process(BoltStateMachineV1.java:92)
    at org.neo4j.bolt.messaging.BoltRequestMessageReader.lambda$doRead$1(BoltRequestMessageReader.java:89)
    at org.neo4j.bolt.runtime.MetricsReportingBoltConnection.lambda$enqueue$0(MetricsReportingBoltConnection.java:68)
    at org.neo4j.bolt.runtime.DefaultBoltConnection.processNextBatch(DefaultBoltConnection.java:191)
    at org.neo4j.bolt.runtime.MetricsReportingBoltConnection.processNextBatch(MetricsReportingBoltConnection.java:86)
    at org.neo4j.bolt.runtime.DefaultBoltConnection.processNextBatch(DefaultBoltConnection.java:139)
    at org.neo4j.bolt.runtime.ExecutorBoltScheduler.executeBatch(ExecutorBoltScheduler.java:171)
    at org.neo4j.bolt.runtime.ExecutorBoltScheduler.lambda$scheduleBatchOrHandleError$2(ExecutorBoltScheduler.java:154)
    at java.util.concurrent.CompletableFuture$AsyncSupply.run(CompletableFuture.java:1604)
    at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
    at java.lang.Thread.run(Thread.java:748)
Caused by: org.neo4j.internal.kernel.api.exceptions.EntityNotFoundException: Unable to load RELATIONSHIP with id 147.
    at org.neo4j.kernel.impl.storageengine.impl.recordstorage.RecordStorageReader.relationshipVisit(RecordStorageReader.java:258)
    at org.neo4j.kernel.impl.coreapi.TxStateTransactionDataSnapshot.relationship(TxStateTransactionDataSnapshot.java:349)
    ... 35 more

The specific transactions that fail with this error seem to be ones that include a deleted relationship. And that relationship was itself deleted by a 'before' trigger, if that matters.

How to Reproduce the Problem

I have not yet been able to reduce the problem to a straightforward procedure to reproduce it. It seems to be a complex interaction between triggers.

Steps (Mandatory)

  1. Add any trigger to phase 'after'
  2. Attempt to execute a transaction that delete a relationship from a 'before' trigger.

Specifications (Mandatory)

Currently used versions

Versions

neo-technology-build-agent commented 2 years ago

Comment by kraney Wednesday Mar 18, 2020 at 02:21 GMT

Correction - that should be version 3.5.16-enterprise

neo-technology-build-agent commented 2 years ago

Comment by vga91 Monday Jun 28, 2021 at 07:57 GMT

@kraney With the latest Apoc version there is a phase: "afterAsync" to prevent some Transaction errors like this. Please, could you try with this?

neo-technology-build-agent commented 2 years ago

Comment by magaton Tuesday Jan 18, 2022 at 19:39 GMT

I am having the same problem. I have the trigger in before phase that is listening to removed property on a deleted relationship (HAS_BOSS). And I want to rename relationships (not the one I am listening to) but the incoming HAS_BOSS relationship of the start node (Client)

     CALL apoc.trigger.add('hasBossStartDateRemoved', 
    "UNWIND $removedRelationshipProperties['startDate'] AS map
      WITH map.relationship AS rel, map.old AS startDate WHERE type(map.relationship) = 'HAS_BOSS'
      WITH rel, startNode(rel) AS client, startDate
      MATCH (client)<-[ohs:HAS_BOSS]-(orphan:Client) 
      WITH ohs
      CALL apoc.refactor.setType(ohs, 'HAD_BOSS') YIELD output
      RETURN count(*)")
  This blows with `Getting deleted relationship data should have been covered by the tx state`
  but the strange thing is that all updates/deletes actuallysucceed.