summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* beta->delta conversion rate is naturally limitedMatthias Radestock2014-03-041-9/+11
| | | | | which means we don't need an explicit upper bound, which in turn would have required some interaction with q_process.
* increase IO_BATCH_SIZEMatthias Radestock2014-03-041-7/+7
| | | | | | | | ...so that beta->gamma conversion can keep up with alpha->beta conversion. Also correct comment - beta->gamma conversion can happen in a lot more places than 'publish'
* reduce_memory_use now only does one thingMatthias Radestock2014-03-031-37/+12
|
* propagate BQ API changeMatthias Radestock2014-03-032-2/+9
|
* explain (reduced) impact of IO_BATCH_SIZEMatthias Radestock2014-03-031-14/+7
|
* decouple syncing from pagingMatthias Radestock2014-03-032-22/+14
|
* merge in defaultMatthias Radestock2014-03-031-2/+3
|\
| * merge bug26049 into defaultMatthias Radestock2014-03-030-0/+0
| |\
| * | clamp duration to infinity when all rates are very lowMatthias Radestock2014-03-031-2/+3
| |/ | | | | | | | | ...so that the memory monitor excludes the queue from avg duration calculations, which would otherwise get massively skewed.
* | limit alpha->beta conversion through msg store credit flowMatthias Radestock2014-03-032-16/+23
|/ | | | | | | we also bump the IO_BATCH_SIZE; it's only used to limit beta->gamma conversion, which is a fairly cheap operation (it only touches qi, which has a very compact on-disk representation), and we want to make sure that such conversion can keep up with the alpha->beta conversion.
* Merge bug26042Simon MacMullen2014-02-283-16/+22
|\
| * Make mirrored queue logging configurable.Simon MacMullen2014-02-273-16/+22
|/
* Merge bug26027Simon MacMullen2014-02-260-0/+0
|\
* \ Merge in defaultSimon MacMullen2014-02-262-9/+14
|\ \ | |/
| * stable to defaultSimon MacMullen2014-02-2660-1845/+2774
| |\
| | * Merge bug26039Simon MacMullen2014-02-261-8/+2
| | |\
| | | * Don't quote node atoms, it makes the resulting string harder to use in shell ↵Simon MacMullen2014-02-261-8/+2
| | |/ | | | | | | | | | scripts.
| * | Merge bug 26038Simon MacMullen2014-02-261-1/+12
| |\ \
| | * | If we abandon autoheal while in winner_waiting then let the losing nodes ↵Simon MacMullen2014-02-261-1/+12
| |/ / | | | | | | | | | know they can carry on.
* | | Update comment.Simon MacMullen2014-02-261-4/+7
| | |
* | | Eliminate the node_stopped message, since it is possible that a badly-timed ↵Simon MacMullen2014-02-262-22/+22
| |/ |/| | | | | stop_app could lead to us missing it. Instead just go based on whether the rabbit stops - if it stops for any reason other than autoheal, we just send it a message it will ignore and continue.
* | Refactor: distinction between handle_dead_rabbit and ↵Simon MacMullen2014-02-261-13/+9
| | | | | | | | handle_dead_rabbit_state is not very useful.
* | Merge bug 26004Simon MacMullen2014-02-241-20/+26
|\ \
| * | trigger rate calculation in all the right placesMatthias Radestock2014-02-071-20/+26
| | |
* | | Merge bug25994Simon MacMullen2014-02-241-2/+11
|\ \ \
| * | | Sort by descending priority order, and add a comment.Simon MacMullen2014-02-241-2/+11
| | | |
| * | | always prioritise acks (and resume)Matthias Radestock2014-02-241-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The former, so that we preserve ack ordering which, while of no semantic relevance, is good for performance since it allows us to optimise for the common acks-in-order case. The latter since it unblocks a queue from delivering messages and thus is conducive to keeping queue lengths short. We prioritise acks higher than most other messages since they are the only high-frequency message which is guaranteed to free up resources. Prioritising ack and resume all the time is safe since flow control places an upper bound on how many of these we can get, thus preventing starvation of senders of lower-priority messages.
* | | | merge stable into defaultMatthias Radestock2014-02-241-32/+44
|\ \ \ \ | |/ / / |/| | / | | |/ | |/|
| * | merge bug26013 into stableMatthias Radestock2014-02-241-32/+44
| |\ \
| | * | improve clarity and correctnessMatthias Radestock2014-02-201-30/+38
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - introduce blocked_by_alarm state predicate which makes it easier / more obvious to detect the transitions to/from - do not send 'unblocked' when the client doesn't have the capability - ditch superfluous true/false return from maybe_send_unblocked (now send_unblocked) - detect the transition into blocked_by_alarm that can result from handling conserve_resources (the original purpose of the bug)
| | * | Eliminate a compiler warningMichael Klishin2014-02-191-1/+1
| | | |
| | * | Send connection.unblocked when connection is no longer blocked or all alarms ↵Michael Klishin2014-02-191-1/+4
| | | | | | | | | | | | | | | | clear
| | * | Make connection.[un]blocked independent from flow controlMichael Klishin2014-02-191-16/+17
| | | |
* | | | Bug 25939 broke AMQP 1.0; fix it.Simon MacMullen2014-02-191-1/+1
| | | |
* | | | stable to defaultSimon MacMullen2014-02-1960-1809/+2733
|\ \ \ \ | |/ / / |/| | |
| * | | Merge bug26022Simon MacMullen2014-02-190-0/+0
| |\ \ \
| * | | | Add LDAP timeout to example configSimon MacMullen2014-02-191-0/+4
| |/ / /
| * | | cosmeticMatthias Radestock2014-02-181-3/+3
| | | |
| * | | merge bug25868 into defaultMatthias Radestock2014-02-180-0/+0
| |\ \ \
| | * \ \ Merge bug25868 (again)Simon MacMullen2014-02-181-5/+4
| | |\ \ \
| | * \ \ \ Merge bug25868Simon MacMullen2014-02-180-0/+0
| | |\ \ \ \
| * | | | | | clearerMatthias Radestock2014-02-181-5/+7
| | |_|/ / / | |/| | | | | | | | | | | | | | | | I think
| * | | | | Only ever return 'flow' if there are no alarms. I think in theory it's not ↵Simon MacMullen2014-02-181-5/+4
| |/ / / / | | | | | | | | | | | | | | | possible for there to be an alarm when in flow control, but this is more correct and no less clear.
| * | | | OopsSimon MacMullen2014-02-171-1/+0
| | | | |
| * | | | ...but only do that if we are not also alarmed.Simon MacMullen2014-02-171-2/+4
| | | | |
| * | | | Show state as 'flow' if we are currently flow blocked and have been so for ↵Simon MacMullen2014-02-172-6/+8
| |/ / / | | | | | | | | | | | | more than 5s.
| * | | Merge bug25627 (again)Simon MacMullen2014-02-1748-1097/+1628
| |\ \ \
| | * \ \ Merge bug25627Simon MacMullen2014-02-171-1/+7
| | |\ \ \
| | * \ \ \ stable to defaultSimon MacMullen2014-02-172-8/+10
| | |\ \ \ \ | | | | |/ / | | | |/| |
| | * | | | merge bug25853 into defaultMatthias Radestock2014-02-133-49/+66
| | |\ \ \ \
View Lorry Controller Status