[
https://issues.apache.org/jira/browse/DISPATCH-1406?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16924200#comment-16924200
]
Ken Giusti commented on DISPATCH-1406:
--------------------------------------
The root cause has been identified:
The problem occurs when a router's defaultTreatment is set to "unavailable",
and a message traveling between routers (over the inter-router link) can no
longer be forwarded (target address is removed from the routing table - think
subscriber lost).
In this situation the router incorrectly avoids replenishing credit on the
inter-router link, which causes eventual credit starvation (after linkCapacity
is exhausted). The router never recovers credit in this situation and
forwarding between the affected routers stalls.
> Inter-router link stall on receive client failover
> --------------------------------------------------
>
> Key: DISPATCH-1406
> URL: https://issues.apache.org/jira/browse/DISPATCH-1406
> Project: Qpid Dispatch
> Issue Type: Bug
> Components: Router Node
> Affects Versions: 1.7.0, 1.8.0
> Reporter: Ken Giusti
> Assignee: Ken Giusti
> Priority: Major
> Fix For: 1.9.0
>
>
> I have a test that simulates a receive client fail-over that results on an
> inter-router link "stalling" - no longer able to forward over the link. From
> what it appears there is 250 (inter-router link capacity) messages that are
> sent over the link but are not accounted for on the receiver end.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
