| Commit message (Collapse) | Author | Age | Files | Lines | ||
|---|---|---|---|---|---|---|
| ... | ||||||
| | | | | * | | | Slaves must join gm before updating amqqueue records, otherwise messages | Tim Watson | 2012-08-29 | 1 | -15/+17 | |
| | | | |/ / / | | | | | | | | | | | | | | | | | | | | | | | | | received from publishers will become stuck due to diverging sender queue state in the slave failing to correlate with incoming messages | |||||
| | | * | | | | Terseness.bug25000 | Simon MacMullen | 2012-09-06 | 1 | -1/+1 | |
| | | | | | | | ||||||
| | | * | | | | Not sure it's worth explaining that, the explanation is not clear and I ↵ | Simon MacMullen | 2012-09-06 | 1 | -3/+1 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | can't think how to improve it. | |||||
| | | * | | | | emit warning when starting detached rabbit | Francesco Mazzoli | 2012-09-06 | 1 | -1/+3 | |
| | | | | | | | ||||||
| | | * | | | | don't record pid when using -detached | Francesco Mazzoli | 2012-09-05 | 2 | -5/+15 | |
| | | | | | | | ||||||
| | | * | | | | rename branch | Francesco Mazzoli | 2012-08-24 | 0 | -0/+0 | |
| | | | | | | | ||||||
| | | * | | | | document the fact that -detached causes the wrong pid to be recorded25000 | Francesco Mazzoli | 2012-08-24 | 1 | -1/+3 | |
| | | |/ / / | ||||||
| | | * | | | tiny refactor: rename fun and clarify comment | Matthias Radestock | 2012-08-24 | 2 | -13/+13 | |
| | | | | | | ||||||
| | * | | | | Try to be a bit more defensive before looping.bug25114 | Simon MacMullen | 2012-08-23 | 1 | -2/+11 | |
| | |/ / / | ||||||
| | * | | | merge bug25112 into default | Matthias Radestock | 2012-08-23 | 0 | -0/+0 | |
| | |\ \ \ | | | |/ | | |/| | ||||||
| | * | | | non-zero timeout when idle for the slave as wellbug25112 | Francesco Mazzoli | 2012-08-23 | 1 | -3/+3 | |
| | |/ / | ||||||
| | * | | cosmetic | Matthias Radestock | 2012-08-23 | 1 | -25/+18 | |
| | | | | ||||||
| * | | | add mochinum.erl, which is needed by mochijson2.erl | Francesco Mazzoli | 2012-09-10 | 2 | -2/+359 | |
| | |/ |/| | ||||||
| * | | accept parameters in JSON format | Francesco Mazzoli | 2012-08-23 | 1 | -60/+10 | |
| | | | ||||||
| * | | re-introduced wrapper functions to strip and add the `struct' tags | Francesco Mazzoli | 2012-08-23 | 1 | -1/+21 | |
| | | | | | | | | | | | We can live with the ambiguity of the empty list, since we only use the term->json function to display the JSON back to the users | |||||
| * | | put result in `ok' | Francesco Mazzoli | 2012-08-23 | 1 | -4/+4 | |
| | | | ||||||
| * | | wrappers around mochijson2 to catch exceptions | Francesco Mazzoli | 2012-08-23 | 1 | -0/+20 | |
| | | | ||||||
| * | | remove the wrapper functions around mochijson2 | Francesco Mazzoli | 2012-08-21 | 1 | -21/+0 | |
| | | | | | | | | | | | | | In the end it's better to keep the `struct' tag - empty lists are ambiguous. I was mislead by the fact that the management plugin removes the outer `struct'. | |||||
| * | | add mochijson2 and license, plus some helper functions | Francesco Mazzoli | 2012-08-21 | 3 | -0/+923 | |
| |/ | | | | | The helper functions are there to remove the tags that mochijson2 uses to distinguish json objects. | |||||
| * | cosmetic - save some lines | Matthias Radestock | 2012-08-20 | 1 | -31/+14 | |
| | | ||||||
| * | Merged bug25104 into default | Emile Joubert | 2012-08-20 | 1 | -3/+1 | |
| |\ | ||||||
| | * | So strip all that stuff out.bug25104 | Simon MacMullen | 2012-08-17 | 1 | -28/+1 | |
| | | | ||||||
| | * | I was thinking that which_applications needed to come first so that the | Simon MacMullen | 2012-08-17 | 1 | -2/+2 | |
| | | | | | | | | | | | | | process would be gone by the time we got to is_process_alive. But if we swap them round the which_applications check will return false in the shutdown case anyway. And of course is_process_alive/1 is rather faster. | |||||
| | * | Gotta love those timeouts. | Simon MacMullen | 2012-08-17 | 1 | -1/+1 | |
| | | | ||||||
| | * | Fix detection of mirror removal at clean shutdown | Simon MacMullen | 2012-08-16 | 1 | -3/+28 | |
| | | | ||||||
| * | | merge bug25071 | Francesco Mazzoli | 2012-08-17 | 18 | -177/+193 | |
| |\ \ | ||||||
| | * \ | merge bug25105 into default | Tim Watson | 2012-08-17 | 1 | -1/+1 | |
| | |\ \ | | |/ | |/| | ||||||
| | | * | Disable parallel build (equivalent to supplying build.jobs=1 on the port ↵bug25105 | Steve Powell | 2012-08-17 | 1 | -1/+1 | |
| | |/ | | | | | | | command). | |||||
| | * | Oops, this got missed in bug 25048 - slave_pids is no longer in both lists ↵ | Simon MacMullen | 2012-08-15 | 1 | -1/+1 | |
| | | | | | | | | | so there's no need to remove it. | |||||
| | * | Merged bug25097 into default | Emile Joubert | 2012-08-13 | 3 | -34/+52 | |
| | |\ | ||||||
| | | * | Merged bug25094 into default | Emile Joubert | 2012-08-13 | 2 | -33/+51 | |
| | | |\ | ||||||
| | | | * | re-assess coverage suspension in testsbug25094 | Matthias Radestock | 2012-08-11 | 1 | -21/+37 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | I took out all the coverage suspensions and then gradually added them back in until the tests passed. The end result is - reduced scope of coverage suspension in some cases - new instances of coverage suspension due to the changes in this bug - some useful helper functions for coverage suspension | |||||
| | | | * | be a bit more selective about which nodes we disconnect from | Matthias Radestock | 2012-08-11 | 1 | -11/+12 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | ...in case we one day relax the restriction that all erlang distribution nodes must be part of the same cluster. - force=false -> all nodes we are clustered with - force=true -> all nodes we are connected to Forceful resets are employed to cope with various conditions in which the cluster has gotten into a bad state, so its best not to rely on mnesia's knowledge of the cluster state. Also: - some drive-by refactoring - nodes() never includes node(), so no need to filter the former | |||||
| | | | * | just connect from all (visible) nodes instead | Matthias Radestock | 2012-08-10 | 1 | -14/+7 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | That way we don't rely on our current (possibly imperfect) knowledge of what the node was clustered with in mnesia. This is safe because, as per the comment, only mnesia cluster nodes should be part of the erlang distributed system. Note that rabbitmqctl et al run as *hidden* nodes. | |||||
| | | | * | make unclustering work for ram nodes | Matthias Radestock | 2012-08-10 | 1 | -10/+18 | |
| | | | | | | | | | | | | | | | | | | | | | - leave the cluster properly when force=false - disconnect from nodes we know about when force=true | |||||
| | | * | | fix a long-outstanding unresolved type reference | Matthias Radestock | 2012-08-12 | 1 | -1/+1 | |
| | | | | | | | | | | | | | | | | | | | | | this was actually completely wrong; the return value is that of *gen_server2*:start_link | |||||
| | * | | | don't re-schedule timer when the new timer would go off <1ms soonerbug25097 | Matthias Radestock | 2012-08-13 | 1 | -1/+1 | |
| | | | | | | | | | | | | | | | | | ...since timer granularity is 1ms | |||||
| | * | | | prompt expiry of requeued messages | Matthias Radestock | 2012-08-13 | 1 | -13/+18 | |
| | | | | | | | | | | | | | | | | | which may require moving the timer expiry forward | |||||
| | * | | | merge default into bug 25097 | Matthias Radestock | 2012-08-12 | 6 | -36/+47 | |
| | |\ \ \ | | |/ / | |/| | | ||||||
| | | * | | propagate API change | Matthias Radestock | 2012-08-12 | 4 | -13/+15 | |
| | | | | | ||||||
| | | * | | more correctness | Matthias Radestock | 2012-08-12 | 1 | -3/+6 | |
| | | | | | ||||||
| | | * | | oops; caught out by daft Erlang scoping again | Matthias Radestock | 2012-08-12 | 1 | -1/+1 | |
| | | | | | ||||||
| | | * | | schedule message expiry timer based on expiry time of queue head | Matthias Radestock | 2012-08-12 | 2 | -22/+28 | |
| | | | | | ||||||
| | * | | | merge bug25096 into default | Matthias Radestock | 2012-08-12 | 1 | -8/+10 | |
| | |\ \ \ | | |/ / | ||||||
| | | * | | don't rely on evil erlang scoping | Matthias Radestock | 2012-08-12 | 1 | -8/+10 | |
| | | | | | ||||||
| | | * | | merge bug25096 into default | Matthias Radestock | 2012-08-11 | 2 | -13/+14 | |
| | | |\ \ | ||||||
| | * | | | | oopsbug25096 | Matthias Radestock | 2012-08-12 | 1 | -2/+2 | |
| | | |/ / | |/| | | ||||||
| | * | | | test 'rabbitmqctl eval' | Matthias Radestock | 2012-08-11 | 2 | -13/+14 | |
| | |/ / | | | | | | | | | | and refactor the error handling for it | |||||
| | * | | test more branches of 'rabbitmqctl set_vm_memory_high_watermark' | Matthias Radestock | 2012-08-11 | 1 | -0/+3 | |
| | |/ | | | | | | | and leave the HWM unchanged! | |||||
| | * | cosmetic | Matthias Radestock | 2012-08-10 | 1 | -3/+4 | |
| | | | ||||||
 |
index : delta/rabbitmq-server.git | |
| hg.rabbitmq.com: Obsolete (use rabbitmq-server-git) |
| summaryrefslogtreecommitdiff |