feat(new_metrics): support server_stat command showing some important server-level metrics (part 3)

[empiredan]

#2382

Add metrics related to the number of RocksDB as the 3rd part to be shown by
server_stat command.

[acelyc111-bot]

Review: server_stat metrics expansion

Summary: Adds 3 new metrics to the command output: , , and . Also expands entity filter to include type.

What's good:

• Clean, consistent implementation — follows existing patterns for struct fields, JSON serialization, and aggregation
• Proper use of for replica-level metrics being aggregated across nodes
• Minimal, focused change (1 file, +45/-14)

Notes:

• The is fetched from the entity but the other two compaction metrics come from entity — this is correct since block cache is shared at server level while compaction tasks are per-replica, but worth documenting
• No unit tests added — consider adding a test for the JSON serialization roundtrip if there's an existing test file for this

Verdict: ✅ Approve — Looks ready to merge. Clean implementation, follows existing conventions.

[acelyc111-bot]

Review: server_stat metrics expansion

Summary: Adds 3 new metrics to the server_stat command output: rdb_block_cache_mem_usage_bytes, rdb_manual_compact_queued_tasks, and rdb_manual_compact_running_tasks. Also expands entity filter to include replica type.

What's good:

• Clean, consistent implementation — follows existing patterns for struct fields, JSON serialization, and aggregation
• Proper use of create_sums for replica-level metrics being aggregated across nodes
• Minimal, focused change (1 file, +45/-14)

Notes:

• The block cache metric is fetched from the server entity while compaction metrics come from replica entity — correct design, but could use a comment explaining the distinction
• No unit tests added — consider adding a test for the JSON serialization roundtrip if there's an existing test file for this

Verdict: ✅ Approve — Looks ready to merge. Clean implementation, follows existing conventions.