[#MULE-242] Refactor Inbound and Response aggregators to support pluggable persistence strategies

Details

Type: Improvement
Status: Reopened
Priority: Major
Resolution: Unresolved
Affects Version/s: 1.0
Fix Version/s: Product Backlog
Component/s: Core: Queues (SEDA) / Persistence, Core: Routing / Filters
Labels:
- clustering

User impact:
Medium
Similar Issues:
~~MULE-4100~~ Support failOnTimeout option for all inbound aggregator routers
~~MULE-3179~~ Inbound Aggregators should also have a timeout feature
MULE-4112 Collection aggregator router seems to require the inbound-endpoint to be synchronous
MULE-2745 Support configuration of persistence strategies (and queueManager?) in 2.x
~~MULE-3925~~ Add failOnTimeout flag to the inbound Aggregator.
~~MULE-2686~~ Customizable Reply & Response Aggregate ID for AbstractResponseRouter
~~MULE-4611~~ Add support for pluggable transformer finders
MULE-5020 Support other messaging patterns for aggregation
~~MULE-949~~ Provide cancellation strategy for aggregation
~~MULE-502~~ Connection strategies to work on inbound and outbound endpoints (asynchronic scenario)

Issue Links

This issue relates to:
~~MULE-1074~~ Harden AbstractResponseAggregator against memory leaks (real and potential)
~~MULE-1300~~ IdempotentReceiver has guaranteed memory leak
~~MULE-841~~ Inconsistent/duplicate class hierarchy with Correlation/ResponseCorrelationAggregators

Activity

Ascending order - Click to sort in descending order

Hide

Permalink

Ross Mason added a comment - 21/Feb/06 05:39 AM

This is supported in Mule 2.0

Show

Ross Mason added a comment - 21/Feb/06 05:39 AM This is supported in Mule 2.0

Hide

Permalink

Travis Carlson added a comment - 15/Aug/06 01:13 PM

This will be possible in the future by using the BPM Connector (~~MULE-899~~)

Show

Travis Carlson added a comment - 15/Aug/06 01:13 PM This will be possible in the future by using the BPM Connector (~~MULE-899~~)

Hide

Permalink

Holger Hoffstaette added a comment - 02/Oct/06 05:30 AM

Travis, can you say a bit more about how this might work out? I think some basic lightweight pluggable persistence strategy could be useful, memory being the default and maybe jdbm (persistent hashmap) or jdbc as simple alternatives.

Show

Holger Hoffstaette added a comment - 02/Oct/06 05:30 AM Travis, can you say a bit more about how this might work out? I think some basic lightweight pluggable persistence strategy could be useful, memory being the default and maybe jdbm (persistent hashmap) or jdbc as simple alternatives.

Hide

Permalink

Travis Carlson added a comment - 03/Oct/06 04:27 PM

Well, once you've got an (embedded or external) BPMS, you could define a process where each incoming message is stored as a process variable or updates a single composite process variable. If persistence is turned on, the process state and variables get persisted on each update. This could be heavyweight or lightweight depeding on which BPMS and which persistence mechanism is used.

Of course, Mule having its own pluggable persistence strategy independent of BPM would be a fine thing as well, the more functionality + options, the merrier!

Show

Travis Carlson added a comment - 03/Oct/06 04:27 PM Well, once you've got an (embedded or external) BPMS, you could define a process where each incoming message is stored as a process variable or updates a single composite process variable. If persistence is turned on, the process state and variables get persisted on each update. This could be heavyweight or lightweight depeding on which BPMS and which persistence mechanism is used. Of course, Mule having its own pluggable persistence strategy independent of BPM would be a fine thing as well, the more functionality + options, the merrier!

Hide

Permalink

Holger Hoffstaette added a comment - 04/Oct/06 05:38 AM

Reopening this since it really seems useful to have at least some kind of pluggable strategy. In-memory would not be much work compared to what we have now. This looks similar to MULE-774 and seems perfect for Spaces (several Mules dropping their loads and one picking it up for aggregation)..

Show

Holger Hoffstaette added a comment - 04/Oct/06 05:38 AM Reopening this since it really seems useful to have at least some kind of pluggable strategy. In-memory would not be much work compared to what we have now. This looks similar to MULE-774 and seems perfect for Spaces (several Mules dropping their loads and one picking it up for aggregation)..

People

Assignee:

Unassigned

Reporter:

Guillaume Nodet

Vote (0)

Watch (0)

Dates

Created:

12/May/05 09:26 AM

Updated:

09/Sep/10 09:08 PM