Existing documentation URL(s)
#28121 and all related pages
What changes are you suggesting?
When using Hyperdrive: You do not have to call client.end() or sql.end() (or similar) - Workers <> Hyperdrive connections will be cleaned up when the request or invocation ends (such as a Workflow or Queue) or Durable Object hiberates/is evicted when idle.
Add a :::note[Create database clients inside your handlers] note that clarifies that you do NOT need to create a database connection in the global scope. Workers do not allow [I/O context](link to relevant Workers docs] to be used across requests, and Hyperdrive's distributed connection pooling solves for connection startup/cold connection latency. Using a driver-level pool in the global script scope will leave you with invalid connections that result in failed queries and hard errors.
Fix all examples to comply.
Additional information
No response
Existing documentation URL(s)
#28121 and all related pages
What changes are you suggesting?
When using Hyperdrive: You do not have to call
client.end()orsql.end()(or similar) - Workers <> Hyperdrive connections will be cleaned up when the request or invocation ends (such as a Workflow or Queue) or Durable Object hiberates/is evicted when idle.Add a
:::note[Create database clients inside your handlers]note that clarifies that you do NOT need to create a database connection in the global scope. Workers do not allow [I/O context](link to relevant Workers docs] to be used across requests, and Hyperdrive's distributed connection pooling solves for connection startup/cold connection latency. Using a driver-level pool in the global script scope will leave you with invalid connections that result in failed queries and hard errors.Fix all examples to comply.
Additional information
No response