Ignore already commited transaction during connector metrics collection

[lukasz-stec]

Description

The connector metrics collection can fail if the
transaction was committed after the check to see if the transaction is active. This is expected, and we should not log an error in this case.

Avoids logs like

2026-04-17T11:39:14.300Z        ERROR   dispatcher-query-3      io.trino.execution.QueryStateMachine    Error collecting query catalog metadata metrics: 20260417_113913_00033_3bybf
java.lang.IllegalStateException: Current transaction already committed
        at io.trino.transaction.InMemoryTransactionManager$TransactionMetadata.checkOpenTransaction(InMemoryTransactionManager.java:387)
        at io.trino.transaction.InMemoryTransactionManager$TransactionMetadata.getTransactionCatalogMetadata(InMemoryTransactionManager.java:434)
        at java.base/java.util.Optional.map(Optional.java:260)
        at io.trino.transaction.InMemoryTransactionManager.getOptionalCatalogMetadata(InMemoryTransactionManager.java:202)
        at io.trino.transaction.TransactionManager.getRequiredCatalogMetadata(TransactionManager.java:54)
        at io.trino.metadata.MetadataManager.getMetrics(MetadataManager.java:495)
        at io.trino.tracing.TracingMetadata.getMetrics(TracingMetadata.java:308)
        at io.trino.execution.QueryStateMachine.collectCatalogMetadataMetrics(QueryStateMachine.java:421)
        at io.trino.execution.QueryStateMachine.getQueryStats(QueryStateMachine.java:941)
        at io.trino.execution.QueryStateMachine.getQueryInfo(QueryStateMachine.java:695)
        at io.trino.execution.QueryStateMachine.updateQueryInfo(QueryStateMachine.java:1486)
        at io.trino.execution.scheduler.StageManager.lambda$initialize$0(StageManager.java:167)
        at io.trino.execution.StageStateMachine.lambda$addFinalStageInfoListener$0(StageStateMachine.java:220)
        at io.trino.execution.StateMachine.fireStateChangedListener(StateMachine.java:240)
        at io.trino.execution.StateMachine.lambda$fireStateChanged$0(StateMachine.java:232)
        at io.airlift.concurrent.BoundedExecutor.drainQueue(BoundedExecutor.java:79)
        at io.trino.$gen.Trino_testversion____20260417_113846_1.run(Unknown Source)
        at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1090)
        at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:614)
        at java.base/java.lang.Thread.run(Thread.java:1474)

Additional context and related issues

Release notes

( X) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text:

## Section
* Fix some things. ({issue}`issuenumber`)

[findepi]

The connector metrics collection can fail if the transaction was committed

This means that connector metrics collection is racing with query finish.
It might be unavoidable, but maybe it's not.

Was there any prior discussion on the topic?

[lukasz-stec]

The connector metrics collection can fail if the transaction was committed

This means that connector metrics collection is racing with query finish. It might be unavoidable, but maybe it's not.

Was there any prior discussion on the topic?

There was some discussion in #26900. Mostly with you. From what I remember, the race is there because of the collection of final query info, and query finish is not synchronized (in general, not in the Java way).