⭐️
Sentry
Sentry project error-events source.
Description
Reads one project's error events using GET /api/0/projects/{organization}/{project}/events/.
Use this source to analyze error events in an external warehouse. The existing Sentry sink and its DSN authentication are unchanged.
Key Features
- Batch
- Streaming
- Exactly-once
Only BATCH mode and parallelism 1 are supported. The source reads an explicit time window.
The engine-injected failure policy must be FAIL_FAST; continuing other tables is not supported by this single-project source.
It does not discover projects, read traces or attachments, or create an immutable snapshot.
Prerequisites
Create a bearer token with project:read access to the chosen project. A sink DSN cannot read events.
For a regional or self-hosted deployment, set its HTTPS origin in api_base_url. Redirects are rejected; configure the final API origin directly.
TLS certificate verification remains enabled. HTTP is supported only for fixtures with mock_mode=true and the literal dummy token="mock-token".
See the events API and pagination contract.
Reading and Recovery Contract
Provide absolute RFC3339 start_time and end_time with an offset, and start strictly before end.
Both are normalized to UTC and sent on every page. The source sends full=false and sample=false;
content contains the returned summary event, not a full stack trace. page_size maps to Sentry's per_page parameter.
Pagination follows only the rel="next" link whose results="true"; a next link with results="false" completes the read.
Only its cursor is copied into the next request. The configured origin, endpoint and time bounds are never replaced by the server's link.
Missing/malformed pagination, repeated cursors, invalid rows, duplicate JSON object keys, oversized responses and exceeding max_pages fail the job instead of silently truncating it.
HTTP 429, 500, 502, 503, 504 and transport failures are retried within max_retries.
Retry-After seconds and HTTP dates are respected. Delays exceeding 60 seconds fail so the job can be retried later; the connector never retries earlier than the server requests.
Authentication failures and redirects are not retried. Closing the reader aborts in-flight requests and wakes retry waits.
The timeout bounds HTTP I/O and schedules a per-attempt abort. Synchronous operating-system DNS resolution, JSON parsing and retry waits are not covered by a strict total job deadline.
Recovery replays the entire configured window; cursors are not durable snapshot offsets and are not checkpointed. The reader holds the checkpoint lock during the bounded read, so use manageable windows and suitable checkpoint timeouts. Rows already written to a nontransactional sink can be duplicated after a failed attempt. Configure downstream deduplication if needed. Sentry retention, delayed ingestion, event removal and changes while paginating can affect results. No complete-history, gap-free adjacent-window or snapshot-consistency guarantee is made.
Output Schema
The schema is fixed; custom schema is not supported. All fields use STRING to preserve Sentry identifiers and timestamp representations.
| Field | Nullable | Meaning |
|---|---|---|
| event_id | no | eventID from the response |
| group_id | yes | groupID |
| project_id | yes | projectID |
| date_created | yes | dateCreated, unchanged |
| title | yes | Event title |
| message | yes | Event message |
| platform | yes | Event platform |
| content | no | Complete returned event object serialized as JSON |
Missing optional fields become null. Unexpected types fail. JSON numeric values retain decimal precision, but content is not byte-identical to the HTTP response.
Event content can contain personal data: use access-controlled sinks and avoid logging rows.
The token option is masked by SeaTunnel's default parsed-configuration logging rules.
Options
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
| token | STRING | yes | - | Bearer token with project read access |
| organization | STRING | yes | - | Organization ID or slug |
| project | STRING | yes | - | Project ID or slug |
| start_time | STRING | yes | - | Absolute RFC3339 start |
| end_time | STRING | yes | - | Absolute RFC3339 end |
| api_base_url | STRING | no | https://sentry.io | HTTPS origin without a path, query, fragment or trailing slash |
| page_size | INT | no | 100 | Events requested per page, 1–100 |
| max_pages | INT | no | 10000 | Page safety limit, 1–100000; fails when exceeded |
| max_retries | INT | no | 3 | Additional transient attempts, 0–5 |
| retry_delay_ms | INT | no | 1000 | Minimum retry delay, 1–60000 ms |
| request_timeout_ms | INT | no | 30000 | Per-attempt HTTP connection/socket timeout and abort deadline, 1–120000 ms |
| max_response_bytes | INT | no | 8388608 | Uncompressed body limit, 1024–16777216 bytes |
| mock_mode | BOOLEAN | no | false | Permit an HTTP fixture with dummy token only |
Source common options such as plugin_output remain available.
Project names accept letters, digits, underscores and hyphens. query, full, sample, dsn and a custom schema are not source options.
Example
Set the SENTRY_TOKEN and WAREHOUSE_PASSWORD environment variables before starting the job. The unquoted HOCON substitutions below resolve their values when the configuration is loaded.
env {
parallelism = 1
job.mode = "BATCH"
}
source {
Sentry {
plugin_output = "events"
token = ${SENTRY_TOKEN}
organization = "example"
project = "backend"
start_time = "2026-01-01T00:00:00Z"
end_time = "2026-01-02T00:00:00Z"
}
}
sink {
Jdbc {
plugin_input = "events"
url = "jdbc:postgresql://warehouse:5432/analytics"
driver = "org.postgresql.Driver"
user = "writer"
password = ${WAREHOUSE_PASSWORD}
database = "analytics"
table = "sentry_events"
generate_sink_sql = true
}
}
Create the destination table with the eight STRING-compatible columns above and install its JDBC driver. The integration test runs an engine job against a deterministic Sentry API fixture; it does not establish compatibility with every Sentry cloud or self-hosted release.
Changelog
Change Log
| Change | Commit | Version |
|---|---|---|
| [improve] sentry options (#9261) | https://github.com/apache/seatunnel/commit/4a2f3fa915 | 2.3.11 |
| [Feature][Restapi] Allow metrics information to be associated to logical plan nodes (#7786) | https://github.com/apache/seatunnel/commit/6b7c53d03c | 2.3.9 |
[Improve] Remove use SeaTunnelSink::getConsumedType method and mark it as deprecated (#5755) | https://github.com/apache/seatunnel/commit/8de7408100 | 2.3.4 |
| [Improve][build] Give the maven module a human readable name (#4114) | https://github.com/apache/seatunnel/commit/d7cd601051 | 2.3.1 |
| [Improve][Project] Code format with spotless plugin. (#4101) | https://github.com/apache/seatunnel/commit/a2ab166561 | 2.3.1 |
| [Hotfix][OptionRule] Fix option rule about all connectors (#3592) | https://github.com/apache/seatunnel/commit/226dc6a119 | 2.3.0 |
| [Improve][Connector-V2][Sentry] Unified exception for sentry sink connector (#3513) | https://github.com/apache/seatunnel/commit/94b472b806 | 2.3.0 |
| [Connector][Dependency] Add Miss Dependency Cassandra And Change Kudu Plugin Name (#3432) | https://github.com/apache/seatunnel/commit/6ac6a0a0cd | 2.3.0 |
| [Feature][Sentry Sink V2] Add Sentry Sink Option Rules (#3318) | https://github.com/apache/seatunnel/commit/850f483816 | 2.3.0 |
| [Feature][Connector-V2] Add sentry sink connector #2244 (#2584) | https://github.com/apache/seatunnel/commit/9fd40390a7 | 2.2.0-beta |