What happened: All the dashboards in the ClickHouse plugin are not aligned with the native dashboards:
- ClickHouse - Query Analysis
- ClickHouse - Data Analysis
- ClickHouse - Cluster Analysis
- Simple ClickHouse OTel Dashboard
- Advanced ClickHouse Monitoring Dashboard
The queries of the dashboards in the plugin seems not right. It lacks of the clusterAllReplicas(default, merge('system', '...')) so only shows one replica's metric, and may display a different result if the query run against a different replica.
For example, here is the "ClickHouse - Advanced Monitoring / Queries/second" panel in Grafana.
SELECT toStartOfInterval(event_time, INTERVAL $__interval_s SECOND) AS t, avg(ProfileEvent_Query)
FROM merge('system', '^metric_log')
WHERE $__dateFilter(event_date) AND $__timeFilter(event_time)
GROUP BY t
ORDER BY t WITH FILL STEP $__interval_s
What you expected to happen: The Grafana dashboards should reflect the actual metrics, e.g. the queries should align/reflect the ClickHouse cloud native dashboard.
As a compare, there is the Queries/Second panel in the native advanced dashboard form the ClickHouse cloud console.
WITH toDateTimeOrDefault({from:String}, '', now() - {seconds:UInt32}) AS from,
toDateTimeOrDefault({to:String}, '', now()) AS to
SELECT
toStartOfInterval(event_time, INTERVAL {rounding:UInt32} SECOND)::INT AS t,
avg(metric)
FROM (
SELECT event_time, sum(ProfileEvent_Query) AS metric
FROM clusterAllReplicas(default, merge('system', '^metric_log'))
WHERE event_date BETWEEN toDate(from) AND toDate(to) AND event_time BETWEEN from AND to
GROUP BY event_time
)
GROUP BY t
ORDER BY t WITH FILL STEP {rounding:UInt32} SETTINGS skip_unavailable_shards = 1
How to reproduce it (as minimally and precisely as possible):
Add a ClickHouse datasource in Grafana, and import the dashboards.
Screenshots
Anything else we need to know?:
Environment:
- Grafana version: 13.1.0
- Plugin version: 4.15.0 (The current latest version)
- OS Grafana is installed on:
- User OS & Browser: MacOS Chrome 147.0.7727.56
- Others:
What happened: All the dashboards in the ClickHouse plugin are not aligned with the native dashboards:
The queries of the dashboards in the plugin seems not right. It lacks of the
clusterAllReplicas(default, merge('system', '...'))so only shows one replica's metric, and may display a different result if the query run against a different replica.For example, here is the "ClickHouse - Advanced Monitoring / Queries/second" panel in Grafana.
What you expected to happen: The Grafana dashboards should reflect the actual metrics, e.g. the queries should align/reflect the ClickHouse cloud native dashboard.
As a compare, there is the Queries/Second panel in the native advanced dashboard form the ClickHouse cloud console.
WITH toDateTimeOrDefault({from:String}, '', now() - {seconds:UInt32}) AS from, toDateTimeOrDefault({to:String}, '', now()) AS to SELECT toStartOfInterval(event_time, INTERVAL {rounding:UInt32} SECOND)::INT AS t, avg(metric) FROM ( SELECT event_time, sum(ProfileEvent_Query) AS metric FROM clusterAllReplicas(default, merge('system', '^metric_log')) WHERE event_date BETWEEN toDate(from) AND toDate(to) AND event_time BETWEEN from AND to GROUP BY event_time ) GROUP BY t ORDER BY t WITH FILL STEP {rounding:UInt32} SETTINGS skip_unavailable_shards = 1How to reproduce it (as minimally and precisely as possible):
Add a ClickHouse datasource in Grafana, and import the dashboards.
Screenshots
Anything else we need to know?:
Environment: