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
Calling Pool#acquire().timeout() (the vanilla timeout operator) introduces a timeout that can effectively trigger in two cases:
the pool is depleted and cannot allocate more resources: timeout on pending acquire (time waiting for a resource to be released)
the pool is depleted BUT can allocate new resources: timeout on the allocation of one new resource
It appear that in some cases, the resource allocation can take some more or less arbitrary time that we are ok waiting for if we now we'll end up with a resource, but we don't want to wait more than necessary if we know we can't allocate more resources.
i.e. we want a timeout on the pending case only.
The text was updated successfully, but these errors were encountered:
@simonbasle Currently in Reactor Netty we have this setting only on the connection pool, you are not able to change this for a particular acquire operation.
as discussed off-band, we'll keep the acquire(Duration) overload only. global configurations can be put in place easily enough in downstream libraries.
Calling
Pool#acquire().timeout()
(the vanilla timeout operator) introduces a timeout that can effectively trigger in two cases:It appear that in some cases, the resource allocation can take some more or less arbitrary time that we are ok waiting for if we now we'll end up with a resource, but we don't want to wait more than necessary if we know we can't allocate more resources.
i.e. we want a timeout on the pending case only.
The text was updated successfully, but these errors were encountered: