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
Occasionally, an outbound sender cuts off an endpoint and somehow manages to get into a state where it no longer sends any events to that endpoint ever again for the duration of the outbound sender's existence. The outbound sender receives events and queues them, but no events are ever taken off the queue and attempted to be sent. This leads to the queue slowly filling up and then being cleared when the endpoint is cut off.
The graph of the outgoing_queue_depths for the specific url will look like this:
and the delivery_count metric for that url is always 0.
I believe the root cause of this issue lies somewhere either with how we are emptying the queue:
Occasionally, an outbound sender cuts off an endpoint and somehow manages to get into a state where it no longer sends any events to that endpoint ever again for the duration of the outbound sender's existence. The outbound sender receives events and queues them, but no events are ever taken off the queue and attempted to be sent. This leads to the queue slowly filling up and then being cleared when the endpoint is cut off.
The graph of the
outgoing_queue_depths
for the specific url will look like this:and the
delivery_count
metric for that url is always 0.I believe the root cause of this issue lies somewhere either with how we are emptying the queue:
caduceus/outboundSender.go
Line 440 in 55bce1c
Or with the dispatcher:
caduceus/outboundSender.go
Line 447 in 55bce1c
The text was updated successfully, but these errors were encountered: