Ted Ross created PROTON-2440:
--------------------------------
Summary: Configured heartbeats sometimes don't function in
inter-cloud environments
Key: PROTON-2440
URL: https://issues.apache.org/jira/browse/PROTON-2440
Project: Qpid Proton
Issue Type: Bug
Components: proton-c
Affects Versions: proton-c-0.35.0
Reporter: Ted Ross
The context is Qpid Dispatch Routers connected together with one router in Edge
mode and the other in Interior mode. The router modes are probably not
important for this issue. The connecting (client) router is running on a
bare-metal Fedora Linux system in a container under Minikube. The listening
router is running in a container in Microsoft Azure Kubernetes Service. There
are unknown layers of networks and load balancers between the routers.
The problem is that configured heartbeats sometimes don't occur. In other
words, a trace on the connection shows no heartbeat (empty) frames flowing.
Furthermore, when this happens, the connection is never closed due to lack of
heartbeats.
This lack of heartbeat protection often results in half-disconnected
connections. When this happens, the client-side sees a connection error
(usually related to TLS), closes the connection and reconnects with a new
connection. The server-side sees no error and winds up with two live
connections from the client, only one of which can actually carry messages
to/from the client.
Qpid Dispatch can work around this problem by establishing an application-level
keepalive using message delivery.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
