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
{{ message }}
This repository has been archived by the owner on Oct 4, 2023. It is now read-only.
When commands are received for a client that is longpolling, then we need to buffer commands that arrive between when the message is sent to the longpoller, and when they reconnect. For example,
t=0, A connects via longpoll
t=100, message is received by A's long poller, and A's long poller disconnects
t=101, message is received by Incus for A, but is dropped because there is no client for A
t=102, A connects via longpoll
Instead, we need a threshold (preferably in time-units rather than absolute number of messages) buffer:
t=0, A connects via longpoll.
t=100, message is received by A's long poller, and A's long poller disconnects
t=101, message is received by Incus for A, and is placed into a buffer
t=102, A connects via longpoll, and a pending message is dispatched, and A's long poller disconnects.
t=104, A connects via longpoll.
The text was updated successfully, but these errors were encountered:
jacobgreenleaf
changed the title
When long polling we need a request queue so commands aren't lost when sent at a high rate
When long polling we need to buffer commands
Aug 18, 2015
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
When commands are received for a client that is longpolling, then we need to buffer commands that arrive between when the message is sent to the longpoller, and when they reconnect. For example,
Instead, we need a threshold (preferably in time-units rather than absolute number of messages) buffer:
The text was updated successfully, but these errors were encountered: