ClickHouse Sink

No Grafana dashboards for ClickHouse yet

The CDviz Grafana dashboards are currently written for PostgreSQL + TimescaleDB only. Storing events in ClickHouse is supported at the collector level, but no pre-built dashboards or SQL queries targeting ClickHouse are provided.

If your organisation relies on ClickHouse and would like to sponsor the development of ClickHouse-native CDviz dashboards, please get in touch.

Stores CDEvents in ClickHouse using a user-defined INSERT query template.

Configuration

[sinks.clickhouse]
enabled = true
type = "clickhouse"
url = "http://localhost:8123"
database = "default"
user = "default"
password = "secret"
query = """
  INSERT INTO cdevents_lake (id, type, source, subject, predicate, specversion, timestamp, payload)
  VALUES ({id}, {type}, {source}, {subject}, {predicate}, {specversion}, {timestamp}, {payload})
"""

Parameters

Parameter	Type	Required	Description
type	string	yes	Must be "clickhouse"
url	string	yes	ClickHouse HTTP endpoint (e.g. http://host:8123)
database	string	yes	Target database
query	string	yes	INSERT query template — see Placeholders
user	string	no	ClickHouse username
password	string	no	ClickHouse password
enabled	boolean	yes	Enable/disable this sink

Placeholders

The query template supports the following {field} placeholders, replaced at runtime with values extracted from each CDEvent:

Placeholder	Description	Example value
{payload}	Full CDEvent as a JSON string	{"context":{"id":"..."},...}
{id}	Event ID (UUID)	a3e5f...
{timestamp}	Event timestamp (RFC 3339)	2025-10-20T14:30:00Z
{type}	Full CDEvent type string	dev.cdevents.service.deployed.0.1.1
{source}	Event source URI	https://github.com/org/repo
{subject}	Subject segment extracted from type (3rd dot-segment)	service
{predicate}	Predicate segment extracted from type (4th dot-segment)	deployed
{specversion}	CDEvents spec version	0.4.1

Each placeholder is replaced with a bound parameter (?) — values are never interpolated as raw strings.

Table Schema

Create the target table before enabling the sink. A minimal example:

CREATE TABLE IF NOT EXISTS cdevents_lake
(
    id           String,
    type         String,
    source       String,
    subject      String,
    predicate    String,
    specversion  String,
    timestamp    String,  -- RFC 3339; cast with parseDateTime64BestEffort() as needed
    payload      String,  -- full CDEvent JSON
    inserted_at  DateTime64(3) DEFAULT now64(3)
)
ENGINE = MergeTree()
ORDER BY (type, inserted_at);

TIP

timestamp is stored as String because CDEvent timestamps may include timezone offsets that ClickHouse DateTime64 does not accept directly. Cast it in queries when needed:

parseDateTime64BestEffort(timestamp) AS ts

Credentials via Environment Variables

export CDVIZ_COLLECTOR__SINKS__CLICKHOUSE__URL="https://host:8443"
export CDVIZ_COLLECTOR__SINKS__CLICKHOUSE__PASSWORD="secret"

See Also

• Database Sink (PostgreSQL) — the default storage sink
• CDviz Database — PostgreSQL + TimescaleDB schema and hosting options