Fire Arrow Server 2.6.6
Fire Arrow Server 2.6.6 has been released.
- (feature) Optional separate database connection pools keep API requests and background work from starving each other
- (bugfix) Practitioner, organization, and device detail pages open from their directories again
Separate Connection Pools for Requests and Background Work
All database work shared one connection pool. Heavy background work, such as a large CarePlan enrollment, could take every connection, so FHIR requests stalled or failed with 503. Heavy request traffic could likewise hold up background work.
Starting with 2.6.6, the server can give HTTP requests and background work separate pools. This is off by default, and behavior is unchanged unless fire-arrow.datasource.split.enabled is set to true.
When enabling it, account for the extra database connections each instance opens, and size CarePlan materialization against the background pool. Per-pool Hikari metrics are not reported while the split is enabled.
Bug Fixes
Detail pages opened the Home page
On the administration UI served by Fire Arrow Server, opening a practitioner, organization, or device from its directory showed the Home page instead of the record. Detail pages now open correctly and use addresses of the form /admin/practitioners/detail?id=<id>. Links saved in the old /admin/practitioners/<id> form (or the organization and device equivalents) no longer open the record.