altinity-expert-clickhouse-ingestion

name: altinity-expert-clickhouse-ingestion description: Diagnose ClickHouse INSERT performance, batch sizing, part creation patterns, and ingestion bottlenecks. Use for slow inserts and data pipeline issues.

Insert Performance and Ingestion Analysis

Diagnose INSERT performance, batch sizing, part creation patterns, and ingestion bottlenecks.

Quick Diagnostics (Run First)

1. Current Insert Activity

select
    query_id,
    user,
    elapsed,
    formatReadableSize(written_bytes) as written,
    written_rows,
    formatReadableSize(memory_usage) as memory,
    substring(query, 1, 80) as query_preview
from system.processes
where query_kind = 'Insert'
order by elapsed desc
limit 20

2. Recent Insert Performance (Last Hour)

select
    toStartOfFiveMinutes(event_time) as ts,
    count() as insert_count,
    round(avg(query_duration_ms)) as avg_ms,
    round(quantile(0.95)(query_duration_ms)) as p95_ms,
    sum(written_rows) as total_rows,
    formatReadableSize(sum(written_bytes)) as total_bytes
from system.query_log
where type = 'QueryFinish'
  and query_kind = 'Insert'
  and event_time > now() - interval 1 hour
group by ts
order by ts desc
limit 20

3. Part Creation Rate by Table

select
    database,
    table,
    toStartOfMinute(event_time) as minute,
    count() as parts_created,
    round(avg(rows)) as avg_rows_per_part,
    formatReadableSize(avg(size_in_bytes)) as avg_part_size
from system.part_log
where event_type = 'NewPart'
  and event_time > now() - interval 1 hour
group by database, table, minute
order by parts_created desc
limit 30

Red flags:

• parts_created > 60 per minute (> 1/sec) → Batching too small
• avg_rows_per_part < 10000 → Micro-batches, will cause merge pressure

4. Insert vs Merge Balance

select
    database,
    table,
    countIf(event_type = 'NewPart') as new_parts,
    countIf(event_type = 'MergeParts') as merges,
    countIf(event_type = 'MergeParts') - countIf(event_type = 'NewPart') as net_reduction
from system.part_log
where event_time > now() - interval 1 hour
group by database, table
having new_parts > 10
order by new_parts desc
limit 20

If net_reduction negative → Load altinity-expert-clickhouse-merges for merge backlog analysis

Problem-Specific Queries

Slow Inserts Investigation

-- Find slowest inserts
select
    event_time,
    query_id,
    user,
    query_duration_ms,
    written_rows,
    formatReadableSize(written_bytes) as written,
    formatReadableSize(memory_usage) as peak_memory,
    arrayStringConcat(tables, ', ') as tables,
    substring(query, 1, 100) as query_preview
from system.query_log
where type = 'QueryFinish'
  and query_kind = 'Insert'
  and event_date = today()
order by query_duration_ms desc
limit 20

Insert with MV Overhead

When inserts feed materialized views, slow MVs cause insert delays.

-- Find slow MVs during inserts
select
    toStartOfFiveMinutes(qvl.event_time) as ts,
    qvl.view_name,
    count() as trigger_count,
    round(avg(qvl.view_duration_ms)) as avg_mv_ms,
    round(max(qvl.view_duration_ms)) as max_mv_ms,
    sum(qvl.written_rows) as rows_written_by_mv
from system.query_views_log qvl
where qvl.event_time > now() - interval 1 hour
group by ts, qvl.view_name
order by avg_mv_ms desc
limit 20

-- Correlate slow insert with MV breakdown (requires query_id)
select
    view_name,
    view_duration_ms,
    read_rows,
    written_rows,
    status
from system.query_views_log
where query_id = '{query_id}'
order by view_duration_ms desc

Failed Inserts

select
    event_time,
    user,
    exception_code,
    exception,
    substring(query, 1, 150) as query_preview
from system.query_log
where type like 'Exception%'
  and query_kind = 'Insert'
  and event_date = today()
order by event_time desc
limit 30

Common exception codes:

• 241 (MEMORY_LIMIT_EXCEEDED) → Load altinity-expert-clickhouse-memory
• 252 (TOO_MANY_PARTS) → Load altinity-expert-clickhouse-merges
• 319 (UNKNOWN_PACKET_FROM_CLIENT) → Client/network issue

Batch Size Analysis

-- Analyze actual batch sizes being inserted
select
    database,
    table,
    count() as insert_count,
    round(avg(written_rows)) as avg_batch_rows,
    min(written_rows) as min_batch,
    max(written_rows) as max_batch,
    round(quantile(0.5)(written_rows)) as median_batch
from system.query_log
where type = 'QueryFinish'
  and query_kind = 'Insert'
  and event_date = today()
  and written_rows > 0
group by database, table
having insert_count > 10
order by avg_batch_rows asc
limit 20

Recommendations:

• avg_batch_rows < 1000 → Seriously under-batched
• avg_batch_rows < 10000 → Could improve
• avg_batch_rows > 100000 → Good batching
• Ideal: 10K-1M rows per insert

Data Pipeline Patterns

Kafka Engine Ingestion

-- Check Kafka consumer lag (if using Kafka engine)
select
    database,
    name,
    engine,
    total_rows,
    total_bytes
from system.tables
where engine like '%Kafka%'

-- Kafka-related messages in logs
select
    event_time,
    level,
    message
from system.text_log
where logger_name like '%Kafka%'
  and event_time > now() - interval 1 hour
order by event_time desc
limit 50

Buffer Table Flush Patterns

-- Buffer table status
select
    database,
    name,
    total_rows,
    total_bytes
from system.tables
where engine = 'Buffer'

Ad-Hoc Query Guidelines

Required Safeguards

-- Always limit results
limit 100

-- Always time-bound
where event_date = today()
-- or
where event_time > now() - interval 1 hour

-- For query_log, filter by type
where type = 'QueryFinish'  -- completed
-- or
where type like 'Exception%'  -- failed

Useful Filters

-- Filter by table
where has(tables, 'database.table_name')

-- Filter by user
where user = 'producer_app'

-- Filter by insert size
where written_rows > 1000000  -- large inserts
where written_rows < 100      -- micro-batches

Cross-Module Triggers

Key Settings Reference

-- Check current settings for a user/profile
select name, value, changed
from system.settings
where name like '%insert%'
order by name