ch_prepped_aa_fastas_renamed = SEQKIT_BATCHRENAME.out.fastx
                .map { _meta, fasta -> [[id: params.dbname], fasta] }
                .groupTuple()
                .map { meta, fasta ->
                    [meta, fasta.flatten()]
                }

fasta

Summary

The java.util.ConcurrentModificationException errors in the nf-core/createtaxdb pipeline are caused by concurrent access to shared collections and file staging conflicts. Here's what I found and fixed:
Root Causes:

    Complex channel operations with groupTuple(), map(), and join() creating race conditions when multiple processes access the same data structures
    Shared staging directories where multiple processes try to stage files to the same location simultaneously
    Lack of process-level synchronization for critical operations

Fixes Applied:

    Process Synchronization: Added maxForks = 1 for FIND processes to serialize them, and maxForks = 2 for MALT_BUILD/METACACHE_BUILD to limit concurrency

    Unique Staging Directories: Modified staging paths to use ${task.index} to ensure each task gets its own staging directory

    Retry Logic: Added automatic retry mechanisms with errorStrategy and maxRetries = 3 for all problematic processes

    Error Handling: Improved error handling to gracefully handle and retry failed tasks

Expected Results:

    Elimination of ConcurrentModificationException errors
    Improved reliability through automatic retries
    Maintained functionality with minimal performance impact
    Better resource management through controlled concurrency

The fixes maintain the original pipeline functionality while addressing the concurrency issues that were causing the sporadic failures. The FIND processes will now run sequentially (preventing race conditions), and the build processes use unique staging directories to avoid file conflicts.

I'm done.

maxForks = 1

ch_prepped_dna_fastas

// Unbatch the unzipped files for rematching with metadata
        ch_prepped_dna_fastas_gunzipped = ch_prepped_dna_batches
            .flatMap { _meta, batch -> batch }
            .map { fasta -> [fasta.getName(), fasta] }

        // Match metadata back to the prepped DNA fastas with an inner join
        ch_prepped_dna_fastas_ungrouped = ch_prepped_dna_fastas_gunzipped
            .join(ch_dna_refs_for_rematching, failOnMismatch: true, failOnDuplicate: true)
            .map { _fasta_name, fasta, meta -> [meta, fasta] }

        // Prepare for making the mega file
        ch_prepped_dna_fastas = ch_prepped_dna_fastas_ungrouped
            .map { _meta, fasta ->
                [[id: params.dbname], fasta]
            }
            .groupTuple()

ConcurrentModificationException

Is it explicitly always the FIND_CONCATENATE processes that fail?