You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Draining is a pretty nice feature to have since it allows you to avoid a bunch of errors that occur during cluster roles that happen do to abruptly dropped connections.
I envision the server to have a shutdownWithin(duration) which will signal all open sessions to close as soon as it is possible to do so safely or to forcefully break the connection within the specified duration.
For HTTP/1.x this means sending Connection: close headers on pending dispatches (if the prelude hasn't been sent) and for H2 this means sending the GOAWAY frame.
The text was updated successfully, but these errors were encountered:
Is this going to be an HTTP-only feature or a general feature? If it's general, does it suggest the introduction of another Command to trigger state so the HTTP extensions (e.g., Connection: close) are sent?
Indeed, my first order idea I was a Command will be pushed down the pipeline since this is not a HTTP only feature. In my minds eye, the alternative is for stages to register interest in such events, but that seems significantly more complex and I don't see any clear advantages.
Draining is a pretty nice feature to have since it allows you to avoid a bunch of errors that occur during cluster roles that happen do to abruptly dropped connections.
I envision the server to have a
shutdownWithin(duration)
which will signal all open sessions to close as soon as it is possible to do so safely or to forcefully break the connection within the specified duration.For HTTP/1.x this means sending
Connection: close
headers on pending dispatches (if the prelude hasn't been sent) and for H2 this means sending theGOAWAY
frame.The text was updated successfully, but these errors were encountered: