JusDB vs Mafiree

Audits active part counts per table partition to intercept 'Too many parts' write freeze errors before analytical ingestion halts.

# 1. Inspect total active parts and partition pressure in ClickHouse
clickhouse-client --user dbre_admin --password -q "
SELECT 
  database, 
  table, 
  count() AS active_parts, 
  formatReadableSize(sum(bytes_on_disk)) AS disk_size 
FROM system.parts 
WHERE active = 1 
GROUP BY database, table 
HAVING active_parts > 100 
ORDER BY active_parts DESC;"

# 2. Monitor ongoing background merges and merge execution duration
clickhouse-client --user dbre_admin --password -q "
SELECT 
  table, 
  elapsed, 
  progress, 
  num_parts, 
  result_part_name 
FROM system.merges 
ORDER BY elapsed DESC;"

Detects dropped write mutations and identifies tables with extreme tombstone density causing read latency timeouts.

# 1. Inspect dropped mutations and thread pool pending tasks
nodetool tpstats | grep -E "MutationStage|ReadStage|Dropped"

# 2. Check maximum and average tombstone density per read query
nodetool tablestats | grep -E "Keyspace|Table:|Maximum tombstones|Average tombstones"