From 62b700b251111d7c0cb09d2b99ec0bfaa180b54e Mon Sep 17 00:00:00 2001 From: Alan Conway Date: Wed, 1 Jul 2009 13:47:04 +0000 Subject: Fix members joining cluster while cluster is handling client errors. Previously cluster members could abort if a new member joins while existing members are handling a client error. Now if an update offer arrives while an error is in progress, the offering broker retracts the offer and the newcomer must try again. git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@790163 13f79535-47bb-0310-9956-ffa450edef68 --- qpid/cpp/xml/cluster.xml | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) (limited to 'qpid/cpp/xml') diff --git a/qpid/cpp/xml/cluster.xml b/qpid/cpp/xml/cluster.xml index d9ab6e97cb..474c191b0b 100644 --- a/qpid/cpp/xml/cluster.xml +++ b/qpid/cpp/xml/cluster.xml @@ -27,16 +27,24 @@ Qpid extension class to allow clustered brokers to communicate. - + + - + + - + + + + + + + @@ -91,7 +99,7 @@ - attach sessions, create consumers, set flow with normal AMQP cokmmands. - send /reset additional session state with controls below. - send shadow-ready to mark end of shadow update. - - send update-complete when entire update is complete. + - send membership when entire update is complete. --> @@ -166,6 +174,9 @@ + + + -- cgit v1.2.1