fix(nf-tower): ensure final log checkpoint runs after thread interrupt

[rnaidu-seqera]

Summary

• LogsCheckpoint.await() was re-setting the thread interrupt flag after catching InterruptedException, causing subsequent NIO writes in saveFiles() to
  throw ClosedByInterruptException and leave empty nf-*.txt, nf-*.log, and timeline-*.html files in the GCS work directory
• Fixed by changing await() to return a boolean instead of re-setting the flag, and calling Thread.interrupted() in run() to clear the flag before
  saveFiles() executes — ensuring the final upload always succeeds regardless of interrupt source
• Added synchronized(lock) guard so onFlowComplete() cannot interrupt the thread mid-upload

Test plan

• Existing LogsCheckpointTest tests pass (interval configuration via default, env var, and config file)
• New unit test 'should perform final saveFiles when interrupted mid-sleep' passes — verifies handler.saveFiles() is called exactly once whenonFlowComplete() interrupts the thread mid-sleep, and that the thread interrupt flag is clear at the time of the call
• Manually verified fix on Google Batch CE with GCS work directory and cloudcache.enabled = true: nf-*.txt file uploaded to GCS with full content after workflow completion (previously 0 bytes)

[netlify]

✅ Deploy Preview for nextflow-docs-staging canceled.

Name	Link
🔨 Latest commit	5fa9b4e
🔍 Latest deploy log	https://app.netlify.com/projects/nextflow-docs-staging/deploys/69f926bbcc6d9100081cc685

[jorgee]

Reading the description of the PR I have not clear what was causing the issue and why you set Thread.interrupted() in all the cases. Is it not causing issues in the normal execution?

[pditommaso]

I think the main mistake is to use interrupt to signal the termination that was introduced in previous iteration.

nextflow/plugins/nf-tower/src/main/io/seqera/tower/plugin/LogsCheckpoint.groovy

Lines 59 to 61 in 6027327

	synchronized(lock) {
	thread.interrupt()
	}

interrupt should be use to force the thread interruption (that's not the case).

I'd suggest going in the change history, revert that implementation and proper exiting the thread

[jorgee]

I think it was added in #6787. I see it is in stable 25.10.x. But this branch doesn't include #6939. So, maybe the reported problem is caused by the race condition and interrupt is happening during upload. Because, if the thread is interrupted, it should break without calling saveFiles.

@rnaidu-seqera, Is the issue happening in 25.10.4?