[IMPALA-4391] HdfsScanners drops Status on the floor in multiple places

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: Impala 2.8.0
• Fix Version/s: Impala 2.8.0
• Component/s: Backend
• Target Version: Impala 2.8.0
• Assignee: Tim Armstrong
• Reporter: Tim Armstrong
• Created: 28/Oct/16 15:10
• Updated: 18/Nov/16 17:41
• Resolved: 18/Nov/16 17:40

Description

I was playing around with _attribute_((warn_unused_result)) (IMPALA-2615) and found this:

  void AttachPool(MemPool* pool, bool commit_batch) {
    DCHECK(scan_node_->HasRowBatchQueue());
    DCHECK(batch_ != NULL);
    DCHECK(pool != NULL);
    batch_->tuple_data_pool()->AcquireData(pool, false);
    if (commit_batch) CommitRows(0); <== drops Status
  }

This may have been valid at some point in time, but can now leave the scanner in an unexpected state if allocating the new RowBatch fails.

BaseSequenceScanner::ProcessSplit() also drops the AddScanRanges() status.