| Commit message (Collapse) | Author | Age | Files | Lines |
| ... | |
| |
|
|
|
|
|
|
|
|
|
|
|
| |
the virtualhost.
AMQQueue.configure: new method
AMQQueueFactory: use AMQQueue.configure, don't set things directly
SimpleAMQQueue.configure, MockAMQQueue.configure: implement new method from the interface
QueueConfiguration: use VirtualHostConfiguration for default values if they're unset
VirtualHostConfiguration: if a queue doesn't exist, give it a default configuration. Add methods to get default settings for configuration items
QueueConfigurationTest: test case, checks that inheritance and defaults work along with explicitly set values.
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761733 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
| |
and purger threads will stop when the inMemory values are within the correct range.
Merge of r761671 and r761674 from trunk
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761688 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
|
|
|
| |
which will replace TestableMessageStore. Update to BaseTransactionLog/Test to work correctly with transactions
and to fully test that functionality. Updated StoreContext to know when it is in a transaction as relying on a payload being set is not sufficient as that is not set when running with the MessageMemoryStore and so transactional testing in the BTLT was not correct.
Update to Virtualhost to correctly set the RoutingTable when the specified TransactionLog is wrapped in a BaseTransactionLog.
Merged from r761670 on trunk
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761685 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
message references so that the underlying TransactionLog need not worry about that.
Updated MemoryMS to use this even to ensure that the code is exercised.
To ensure that the new BaseTransactionLog was correctly used when used by a TransactionLog. The configure() method now returns an Object(TransactionLog) that is the newly configured TL. Existing tests and code where the original TL reference was used have
been changed to use the output of the configure() call.
NOTE: the return type should be changed to TransactionLog but until we have completely split the TransactionLog and RoutingTable implementations then this is not possible.
The implementation also includes a number of items from the Flow To Disk review:
- The old get* Methods have been removed from the TransactionLog interface.
- Rollback should now rollback enqueues. (No test provided)
- StoreContext now has enqueue/dequeue methods that track the messageId/Queue pairing
- The linked list per message has been reduced to a link list per message that is enqueued on multiple queues. Messages that exist on only one queue have no additional overhead.
- Optimisation also included to:
Include message delete in 'dequeue transaction' where the message was only ever enqueued on a single queue.
All other message deletes are peformed as part of an asynchrounous commit.
The asynchrounous commit is setup via the StoreContext, which has had some work done to move it towards becomming a Qpid Transaction Object where all operations are performed against rather than going via the TransactionLog.
Merged from r760951 on trunk
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761673 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
| |
checks failover doesn't occur and we know about it in less time than the connectiondelay which would be waited for if we attempted failover.
Merge of r761668 from trunk
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761669 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
|
| |
connection retry but only on the initial connection...
Merged from trunk r759097
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761253 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
|
| |
Relax MessageFactory to allow out of order. Updated test
Merging trunk r760952
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@761251 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
| |
Merge of r758748
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758763 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
|
| |
for a FlowableBaseQueueEntryList to delegate its ac - New Messages on a flowed queue are pushed optimistically pushed to disk, this should potentially be removed and just rel - When space is available messages are loaded in queue order, so in this case Priority order.
Merged revision r758742
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758758 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
| |
Merge of r758733
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758754 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758749 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
load/unload, Updated FQBS to adhere to the thread safety specified by the interface. QueueEntry was updated to return the AMQMessage from the load() to complete the getMessage() interface. As in a flowed state the message may be purged before a reference can be taken. Added new Test QueueEntryImplThreadingTest that should later be run for longer but aims to show that load always returns the message even when unloads are occuring asynchronously.
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758388 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
causing FtD to fail. Annoyingly the tests work as they used the method with the extra functionality.
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@758372 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
the relevant exception message
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@757268 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@757257 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
in r748516
git-svn-id: https://svn.apache.org/repos/asf/qpid/branches/0.5-release@756506 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
KBytes, make size descriptions in other methods consistent
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@755256 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
JMX and Java broker roots
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@755187 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@754958 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@754934 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
of directly in user home directory
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@754807 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
purpose to users
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@754803 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
before we had fully initialised it.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@753253 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
persistent recovery as we do not have a working persistent module.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@753220 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
file from disk then recover from the persistent strore. This approach applies the vhost configuration and then applies the persistent state from the store to those objects rather than recreating them. The new inner classes on VirtualHost are to be removed once we have fully extracted the RoutingTable from the legacy MessageStores as this is the root of the problem. The Store needs to be open to create new durable objects but the current stores must recover their state before new entries are added. So now the persistent state is being loaded on to a broker in a consistent state after it has configured a) its default exchanges and b) loaded the queue config from the config file. Eventually we will only have one location for queue config and all will be right in the world.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@753219 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
can flow to disk transient queues with Apache store
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@752845 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
Simplest solution here is to disable the management interface for these tests. The more complex change to ensure the MC is shut down correctly and port freed during test startup will not be a problem for production. It only occurs here as we start a broker then stop and start a new one immediately.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@752810 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@752529 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@752528 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
will increase our memory usage. Instead add a close method on the BackingStore that is called when the queue closes and clean up the backing then. Also moved the QueueHousekeeping thread stop to before we do any queue closing in the VHost. This will ensure that we are not causing any operations that might inadvertadly load a message, and so prevent the backing file from being deleted. This should of course now not occur as all getMessage() calls have been removed.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751722 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
provided by java.io. Attempting to delete an non-existent file returns false. The caller must handle this correctly. If client provides a null value then the call will throw a NPE which is a valid java response.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751720 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
message to be read in to memory from disk. In all instances the reason was to perform methods that exist on the the QueueEntry. Added accessor to MessageID on QueueEntry. Outstanding getMessage() calls have been left in to perform NO_LOCAL work. Moving Publisher and PublisherClient identifer to the QEI would remove this need.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751718 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
Priority queues.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751714 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751636 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
| |
The fix is a bit of a hack, but I don't see another way to handle it.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@751061 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
null, it would previously have thrown a NPE which was erroneously caught this and declared the delete to have failed. If there is nothing to delete (signified by the Null File object) then the delete should pass.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750946 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
getFileHandle() method so it was attempting to delete a bin rather than the message data. Our Broker Unit tests don't create the right type of ByteBuffer so whilst they passed the buffer.getData().array() failed as array() is not supported in the ByteBuffer types we actually use at runtime.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750876 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
| |
100Meg. Future work would be to have the broker split Xmx between VirtualHost and for the Vhost then to use its housekeeping thread to spread the memory between the current queues. It could of course be clever and give more memory to queues in high use.
Mistype in AMQQueueFactory resulted in it setting the Max Queue Size alert rather than the Max Memory value
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750875 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
null. Used setDeliveredToConsumers() on QEI to know when it is safe to unload message so SAMQQueue does not need to have anything to do with message un/loading. So the FtD work is fully encapuslated by the QEI which delegates memory accounting to its list.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750874 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
prevent the inadvertent loading of a flowed message with a call to getMessage()
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750873 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750871 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
been flowed to disk
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750869 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750868 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
applied to queue.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750867 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
|
| |
The automated test case for this is still being worked out.
However this feature has been tested manually.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750205 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750204 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
|
| |
This includes a the failover test run in a loop.
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750203 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
| |
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@750025 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
fix for QPID-1642/QPID-1643
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@749842 13f79535-47bb-0310-9956-ffa450edef68
|
| |
|
|
|
|
| |
execution exception
git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk@749740 13f79535-47bb-0310-9956-ffa450edef68
|