summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Cosmeticbug24696Emile Joubert2012-04-122-29/+24
|
* queue leases with if_empty & if_unused optionEric Litak2012-04-112-22/+68
|
* oopsMatthias Radestock2012-04-121-1/+1
|
* Merge bug 24769.Simon MacMullen2012-04-122-4/+2
|\
| * tiny space saving refactorMatthias Radestock2012-04-121-3/+1
| |
| * backing queue quickcheck updates for bug20337Emile Joubert2012-04-121-1/+1
| |
* | fix bugbug24769Matthias Radestock2012-04-121-1/+1
| |
* | cosmeticMatthias Radestock2012-04-111-15/+14
| |
* | fix a bugMatthias Radestock2012-04-111-1/+1
| |
* | merge bug24769Matthias Radestock2012-04-1125-473/+803
|\ \ | |/
| * Micro-optimisationSimon MacMullen2012-04-111-1/+1
| |
| * Matthias points out that we don't need to lookup here; we will lookup later ↵Simon MacMullen2012-04-111-3/+1
| | | | | | | | anyway...
| * Merge bug24862Simon MacMullen2012-04-111-0/+7
| |\
| | * Deduplicate routes (needed for sender routing).bug24862Simon MacMullen2012-04-111-1/+2
| | |
| | * Optimise routing for the default exchange.Simon MacMullen2012-04-111-0/+6
| |/
| * Merge bug24725.Simon MacMullen2012-04-116-68/+125
| |\
| | * Merge bug24854Simon MacMullen2012-04-115-67/+124
| | |\
| | | * cosmeticMatthias Radestock2012-04-101-3/+4
| | | |
| | | * optimisation: gm:broadcast earlier so other nodes can get on with workMatthias Radestock2012-04-101-2/+2
| | | |
| | | * merge bug24846 into defaultMatthias Radestock2012-04-103-29/+31
| | | |\
| | | | * Merged bug24833 into defaultEmile Joubert2012-04-102-37/+28
| | | | |\
| | | | | * Merged bug24833 into defaultEmile Joubert2012-04-052-29/+29
| | | | | |\
| | | | * | | Restore rabbit_basic:extract_headers/1Emile Joubert2012-04-101-1/+8
| | | | | |/ | | | | |/|
| | | | * | Fix in map_headers.Francesco Mazzoli2012-04-031-3/+3
| | | | | |
| | | | * | Pushed a change to the Makefile by mistake.Francesco Mazzoli2012-04-031-1/+1
| | | | | |
| | | | * | map_headers instead of {extract, replace}_headers.Francesco Mazzoli2012-04-033-30/+30
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Sadly the final value of the headers depends on whether we have a DLX routing key or not, so the code in rabbit_amqqueue_process:make_dead_letter_msg/4 is slightly ugly. I still think it's better to keep rabbit_basic:map_headers/2 general.
| | | * | | typobug24846Matthias Radestock2012-04-051-1/+1
| | | | | |
| | | * | | dtree API cleanup: handle 'insert' w empty secondary key list correctlyMatthias Radestock2012-04-051-1/+5
| | | | | |
| | | * | | better(?) dtree docsMatthias Radestock2012-04-041-11/+26
| | | | | |
| | | * | | revise nack handling in queue (for dlx)Matthias Radestock2012-04-031-7/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - with the changes introduced in bug 24750, the "lost" message count logged was misleading since it would not include any messages still pending confirmation from other queues. Fix that. - normal queue terminations are not log worthy - splitting log information across two entries loses context, i.e. other log entries might appear inbetween. So now we collate all the information into a single log entry. - saying that messages were "lost" is misleading - we simply don't know what happened to them. Make this clear(er).
| | | * | | fix nack handling in channelMatthias Radestock2012-04-032-6/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a queue dies abnormally we nack all messages sent to that queue and still pending confirms. Any confirms coming in for these messages thereafter - from other queues - are ignored.
| | | * | | make dtree:take/3 cope with absent keysMatthias Radestock2012-04-031-19/+24
| | | | | |
| | * | | | Whoopsy.bug24854Matthew Sackman2012-04-041-1/+1
| | | |_|/ | | |/| |
| * | | | I initially read this as "head 2 is an optimistion of head 3, and head 4 is ↵bug24725Simon MacMullen2012-04-111-1/+1
| | | | | | | | | | | | | | | | | | | | a different case".
| * | | | refactor: shrinkMatthias Radestock2012-04-051-9/+7
| | | | |
| * | | | merge default into bug24725Matthias Radestock2012-04-052-37/+28
| |\ \ \ \ | | |/ / /
| | * | | Merge bug24789Simon MacMullen2012-04-041-7/+2
| | |\ \ \ | | | |/ / | | |/| |
| | | * | Don't invoke userdelbug24789Emile Joubert2012-04-031-7/+2
| | | | |
| | * | | merge bug24845 into defaultMatthias Radestock2012-04-031-13/+11
| | |\ \ \
| | | * | | do not confirm 'immediate' messages that failed to be deliveredbug24845Matthias Radestock2012-04-031-13/+11
| | |/ / / | | | | | | | | | | | | | | | | | | | | in particular non-persistent messages were always being confirmed regardless
| | * | | refactor queue's confirm handling on publishMatthias Radestock2012-04-021-30/+28
| | | |/ | | |/| | | | | | | | | I think it's a bit clearer this way
| * | | handle confirms and dlxMatthias Radestock2012-04-021-12/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When a dlx is configured we fall back to the normal go-via-the-backing-queue route since the dlx logic assumes that any dead-lettered message is present in the bq. Indeed the bq is the only place in which such messages are kept until they have been confirmed by the dead-letter destinations.
| * | | merge default into bug24725Matthias Radestock2012-04-0217-245/+493
| |\ \ \ | | |/ /
| | * | merge bug24843 into defaultMatthias Radestock2012-04-020-0/+0
| | |\ \
| | * | | fix a racebug24843Matthias Radestock2012-04-021-0/+3
| | |/ /
| | * | merge bug24842 into defaultMatthias Radestock2012-04-021-1/+2
| | |\ \
| | | * | fix cluster tests to work again in single-node casebug24842Matthias Radestock2012-04-021-1/+2
| | |/ /
| | * | Merge bug24808.Simon MacMullen2012-04-0225-414/+676
| | |\ \
| | | * \ Merge bug24750.Simon MacMullen2012-04-024-172/+173
| | | |\ \ | | | | |/ | | | |/|
| | | | * Attempt to flesh out the essay slightly more.bug24750Simon MacMullen2012-04-021-7/+15
| | | | |
View Lorry Controller Status