Problem Statement
On console, the crash reporting endpoint is configured per title in the platform's
own developer tooling, outside the application build. Studios shipping multiple
titles must configure a distinct Sentry ingestion URL for each one, in a place that
is easy to get wrong and hard to audit.
In practice crashes from one title end up in another title's project, and the
misconfiguration is only discovered when someone notices unfamiliar crashes in their
dashboard. Because the routing decision lives in platform configuration rather than
in the build, the team that owns the Sentry project is usually not the team that set
the URL.
Misrouting corrupts two projects at once: the receiving project gains noise, and the
originating project silently loses signal. There is no validation at ingestion time
that events arriving on a given URL belong to that project, and every new title
repeats the same error-prone setup step.
Solution Brainstorm
- Accept a single organization-level ingestion URL for console crash reporting, and
route events to the correct project server-side using the title identifier already
present in the crash report.
- Expose the title-to-project mapping as configuration in Sentry, editable by the
team that owns the projects rather than the team that owns platform config.
- As a smaller first step that would catch most of the damage: warn when events
carrying an unexpected title identifier arrive on a project's endpoint.
Product Area
Ingestion
Requested by Saad Khan.
--
View Junior Session [Sentry]
Problem Statement
On console, the crash reporting endpoint is configured per title in the platform's
own developer tooling, outside the application build. Studios shipping multiple
titles must configure a distinct Sentry ingestion URL for each one, in a place that
is easy to get wrong and hard to audit.
In practice crashes from one title end up in another title's project, and the
misconfiguration is only discovered when someone notices unfamiliar crashes in their
dashboard. Because the routing decision lives in platform configuration rather than
in the build, the team that owns the Sentry project is usually not the team that set
the URL.
Misrouting corrupts two projects at once: the receiving project gains noise, and the
originating project silently loses signal. There is no validation at ingestion time
that events arriving on a given URL belong to that project, and every new title
repeats the same error-prone setup step.
Solution Brainstorm
route events to the correct project server-side using the title identifier already
present in the crash report.
team that owns the projects rather than the team that owns platform config.
carrying an unexpected title identifier arrive on a project's endpoint.
Product Area
Ingestion
Requested by Saad Khan.
--
View Junior Session [Sentry]