Which of the following reasons will cause HDFS NameNode to enter safemode (safe read-only mode)? (multiple choice)

Question

Options

• AThe disk space where the metadata of the active and standby NameNodes reside is insufficient.
• BThe number of lost blocks exceeds the threshold.
• CThe missing copy exceeds the threshold.
• DThe damaged copy exceeds the threshold.

• A
  81% (21)
• C
  12% (3)
• D
  8% (2)

Explanation

A is correct because the NameNode must continuously write metadata (fsimage and edit logs) to its local disk - if disk space runs out, it cannot safely persist changes and enters safemode to prevent corruption or data loss.

B is correct because when the percentage of blocks with zero surviving replicas (i.e., completely lost blocks) exceeds the configured threshold (dfs.namenode.safemode.threshold-pct), the cluster cannot guarantee data availability and enters safemode.

C is wrong because a "missing copy" means a block is under-replicated - it still has at least one replica. HDFS handles this quietly in the background by scheduling re-replication, not by entering safemode.

D is wrong for a similar reason - a "damaged copy" (corrupt replica) still leaves other healthy replicas intact. HDFS marks the corrupt replica and re-replicates from a good one without triggering safemode.

Memory tip: Safemode is triggered by existential threats - metadata can't be written (disk full) or data is completely gone (lost blocks). Under-replication and single corrupt replicas are recoverable states HDFS handles silently, so they don't warrant safemode.

No community discussion yet for this question.