WIP: rip out old cluster code, misc.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1424118 13f79535-47bb-0310-9956-ffa450edef68

Having queue state replication between 2 clusters, restarting a broker in both
source+destination clusters sometimes leads to cluster de-sync. No QMF
communication is involved, though symptoms are similar to the bug caused by
missing propagation of QMF errors within a cluster.

The bug is caused by "deliveryCount" in SemanticState::ConsumerImpl
(qpid/broker/SemanticState.cpp) not being replicated to a joining cluster node
during catch-up. When the elder broker in src.cluster sends session.sync() after
sending 5 messages (per --ack 5 in qpid-route), the recently joiner node in
src.cluster does not do so, what leads to the cluster de-sync.

The patch:

 - adds to "consumer-state" method (see xml/cluster.xml file change) to update a new joi-ner a new property deliveryCount
 - updates cluster::Connection::consumerState to send deliveryCount to the method
 - updates cluster::Connection::consumerState to set the received deliveryCount
 - add two methods to broker::SemanticState::ConsumerImpl for getting and setting deliveryCount

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1352588 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1332657 13f79535-47bb-0310-9956-ffa450edef68

Revision r1307582 created a serious degredation in priority queue performance.

It replaced a muti-deque implementation with o(1) complexity for consuming with
a map implementation with o(log(n)) performance.

This revision returns to a mutli-deque algorithm but with the addition
of a FIFO index for fast browsing of acquired and unacquired messages.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1329301 13f79535-47bb-0310-9956-ffa450edef68

Updated priority and lvq queues to keep acquired messages, and supply
them to browsers if requested. This is necessary so replicating
subscriptions can back-up these queue types without message loss.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1307582 13f79535-47bb-0310-9956-ffa450edef68

Only allow brokers that authenticate as the cluster-username to join a cluster.

New broker first connects to  a cluster broker authenticates as the cluster-username
and sends its CPG member ID to the qpid.cluster-credentials exchange.
The cluster broker that subsequently acts as updater verifies that the credentials are
valid before connecting to give the update.

NOTE 1: If you are using an ACL, the cluster-username must be allowed to
publish to the qpid.cluster-credentials exchange. E.g. in your ACL file:

acl allow foo@QPID publish exchange name=qpid.cluster-credentials

NOTE 2: This changes the cluster initialization protocol, you will
need to restart the cluster with all new version brokers.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1210989 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1205467 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1163347 13f79535-47bb-0310-9956-ffa450edef68

- Replicate DTX state to new members joining.
- Use cluster timer for DTX timeouts.
- Incidental: quote nulls in qpid::Msg messages (XIDs often have null characters)

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1161742 13f79535-47bb-0310-9956-ffa450edef68

When sending a large number of messages with nonzero TTLs to a
cluster, overall message throughput drops by around 20-30% compared to
messages with TTL 0.

The previous approach to TTL in the cluster is replaced with a simpler
"cluster clock". Also QueueCleaner is executed in the cluster timer,
and modified to be deterministic in a cluster.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@1136170 13f79535-47bb-0310-9956-ffa450edef68