Sentry to BigQuery: How to Archive Your Data
Move Sentry errors into BigQuery with Airbyte. Why overwrite mode destroys history, the three token scopes you need, and one source per Sentry project.
Replicating Sentry to BigQuery is usually a retention problem wearing an analytics costume. Sentry SaaS keeps error events for 30 or 90 days depending on your plan. After that they are gone, which means any question that spans quarters, correlates errors with releases across a year, or asks whether a class of failure is actually getting worse cannot be answered inside Sentry at all.
BigQuery is where that history lives instead. Airbyte moves it there with no code. But the retention window changes which sync mode is correct, and getting this wrong quietly destroys the archive you are trying to build.
Sync mode is the whole article
Full Refresh Overwrite replaces the destination contents with whatever Sentry currently returns. Against a source with a 30 day retention window, that means every sync deletes everything older than 30 days from your warehouse. You end up with a very expensive copy of a rolling window you already had.
Use an append mode. Incremental append, or full refresh append where a stream does not support incremental, keeps events in BigQuery after Sentry has dropped them. That is the entire point of the pipeline. Configure it once, correctly, at connection setup, because the first overwrite sync after a month of accumulated history is not recoverable.
Token scopes, and the 403 they cause
Create an auth token, or for Sentry SaaS an internal integration with organisation-wide access. Grant read access on organisations, projects, and issues and events. In scope terms that is org:read, event:read and project:read.
A token missing one of these does not fail the connection test in a useful way. It fails the individual stream with an HTTP 403 while the rest of the sync succeeds, so you get a partially populated dataset and a green connection. If a stream is empty and you cannot explain why, check the scopes first.
One project per connection
The source configuration takes an organisation slug and a project slug, and the project setting determines which project the events, issues and project_detail streams read. There is no wildcard. Ten Sentry projects means ten sources, which is tedious to set up and worth doing deliberately: name the destination datasets or table prefixes by project from the start, because renaming them later means rewriting every downstream query.
If you are on a region-specific Sentry domain, set the host name accordingly. Values like us.sentry.io, us2.sentry.io and de.sentry.io are all valid, entered without the scheme or a trailing path. Self-hosted Sentry works the same way with your own hostname. Sentry also applies request and concurrency limits per caller and endpoint, reported in X-Sentry-Rate-Limit headers, so if syncs return HTTP 429 the fix is to lower the number of concurrent workers on the connection. Backfilling ten projects in parallel is the usual way to trip this.
Frequently asked questions
Will Airbyte backfill errors older than my Sentry retention window?
No. Airbyte can only read what Sentry still has. The archive starts on the day your first sync runs, which is a good argument for setting the connection up before you need the history rather than after.
Why is one of my Sentry streams empty?
Most likely a missing token scope returning 403 on that stream alone. Grant org:read, event:read and project:read, then re-run the sync.
Can one connection cover all my Sentry projects?
No. The project slug is part of the source configuration, so you need one source per project.
What sync mode should I use for Sentry?
An append mode, always. Overwrite will delete history from BigQuery every time Sentry ages events out of its retention window.
Get your Sentry data into BigQuery
The value here is not that BigQuery queries errors better than Sentry does. It is that BigQuery still has them in eighteen months. Set append mode at the start, grant all three scopes, and accept one source per project as the cost of doing it properly.
If your team wants full-text search across records rather than SQL aggregation over them, Sentry to Elasticsearch covers the same source indexed for search instead. For another API source with its own permissions model, see Microsoft Dataverse to BigQuery.
Integrate with 600+ apps using Airbyte
Move data from 600+ sources into warehouses, lakes, and beyond. Set up pipelines in minutes with pre-built connectors and the Connector Builder.