-
Notifications
You must be signed in to change notification settings - Fork 2.2k
[BUG] Revisit ReactorNettyHttpResponse::close() #10467
Copy link
Copy link
Closed
Labels
Azure.Coreazure-coreazure-coreClientThis issue points to a problem in the data-plane of the library.This issue points to a problem in the data-plane of the library.HttpClientpillar-reliabilityThe issue is related to reliability, one of our core engineering pillars. (includes stress testing)The issue is related to reliability, one of our core engineering pillars. (includes stress testing)
Metadata
Metadata
Assignees
Labels
Azure.Coreazure-coreazure-coreClientThis issue points to a problem in the data-plane of the library.This issue points to a problem in the data-plane of the library.HttpClientpillar-reliabilityThe issue is related to reliability, one of our core engineering pillars. (includes stress testing)The issue is related to reliability, one of our core engineering pillars. (includes stress testing)
Type
Fields
Give feedbackNo fields configured for issues without a type.
It looks like there are cases where when
ReactorNettyHttpResponse::close()is called the connection is in the disposed state but the underlying body is not drained, which can result in leak warnings from netty.Need to followup with the reactor-netty team to understand the behavior of connection disposal.
This is not impacting the majority of the cases since almost all cases either we subscribe to the body in RestPoxy or by the user when downloading as raw ByteBuf. One case in key-vault revealed this issue where it was not disposing the body inside the policy and was expecting close() to do it