Aiven for ClickHouse
TraceHouse works with Aiven for ClickHouse, but some features operate in degraded mode due to system table restrictions on the managed platform.
Connection
Aiven doesn't return CORS headers, so you need the bundled proxy:
just proxy-start
See Connecting for full setup details.
Known Limitations
Aiven restricts access to several system log tables. The following are not available on Aiven for ClickHouse (as of March 2026):
| Unavailable System Table | Impact |
|---|---|
metric_log | No native server metrics timeseries (CPU, memory, network, disk IO) |
asynchronous_metric_log | No historical async metrics (RAM totals, CPU core counts from log) |
trace_log | No flamegraphs or CPU sampling |
query_thread_log | No per-thread query breakdown |
processors_profile_log | No query pipeline profiling |
opentelemetry_span_log | No distributed tracing spans |
crash_log | No crash/fatal error records |
session_log | No login/logout tracking |
zookeeper_log | No ZooKeeper request log |
backup_log | No backup operation history |
blob_storage_log | No blob storage operation log |
Source: Aiven documentation - System tables
Time Travel - Degraded Mode
Without metric_log, the Time Travel page cannot show true server-level metrics. Instead, it operates in degraded mode:
- Charts display estimated metrics synthesized from
query_logProfileEvents - CPU, memory, network, and disk IO values are derived by distributing each query's totals evenly across its duration
- Gaps between queries appear as zero - the chart only reflects periods where queries were actively running
- Values may undercount actual server usage since background processes (merges, system tasks) that don't appear in
query_logare not captured - Query, merge, and mutation timelines (the bands and tables) work normally since they use
query_logandpart_log
A warning banner and "⚠ Estimated" badge are shown when degraded mode is active.
Features That Work Normally
These features use query_log, part_log, system.parts, and other available tables:
- Database Explorer (tables, parts, columns)
- Merge Tracker (merge history, merge timeline)
- Query Monitor (query history, running queries)
- Cluster Overview (topology, replica status)
- Engine Internals (part inspection, ordering key analysis)
System Log TTL
Aiven sets a 1-hour TTL on all system log tables. This means:
- Query history is limited to the last hour
- Time Travel can only look back ~1 hour
- Merge history is limited to recent events
You can work around this by creating materialized views to persist log data:
CREATE MATERIALIZED VIEW my_query_log
ENGINE = MergeTree
PARTITION BY event_date
ORDER BY event_time
AS SELECT * FROM system.query_log;
Capability Detection
TraceHouse automatically detects which system tables are available on connection. The Capabilities panel (visible in the connection sidebar) shows exactly what's available and what's missing on your Aiven instance.