summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* OopsSimon MacMullen2014-09-091-1/+1
|
* Rename slightly and improve comments.Simon MacMullen2014-09-093-14/+16
|
* OopsSimon MacMullen2014-09-091-1/+1
|
* OopsSimon MacMullen2014-09-091-1/+1
|
* CosmeticSimon MacMullen2014-09-091-7/+6
|
* Introduce the idea that queues can be absent for a reason. The traditional ↵Simon MacMullen2014-09-098-37/+84
| | | | | | absent reason is 'nodedown' and we have a new reason, 'crashed', for when crash recovery has failed and the supervisor has given up. An absent crashed queue is nearly the same as an absent nodedown queue, but we allow for deleting it since it can't be recovered by bringing a node back up. Currently absent crashed queues are not handled properly by mgmt (they appear to still be there); we might defer that to bug 26151.
* Try to be consistent.Simon MacMullen2014-09-092-2/+3
|
* SpecSimon MacMullen2014-09-081-1/+4
|
* Merge in default.Simon MacMullen2014-09-087-34/+59
|\
| * Merge bug26363Simon MacMullen2014-09-081-7/+8
| |\
| | * Filter out nodes where Mnesia is not running.Simon MacMullen2014-09-081-7/+8
| |/
| * stable to defaultSimon MacMullen2014-09-081-2/+4
| |\
| | * Merge bug26362Simon MacMullen2014-09-081-2/+4
| | |\
| | | * Slightly less threatening error description, and add stack traceSimon MacMullen2014-09-081-2/+4
| | |/
| * | We can't assume encoded lengths, so do it all in lists.Simon MacMullen2014-08-221-11/+15
| | |
| * | That's not been relevant since 2.6.0 or something.Simon MacMullen2014-08-221-2/+1
| | |
| * | stable to defaultSimon MacMullen2014-08-2156-845/+2328
| |\ \ | | |/ | |/|
| | * Avoid binary:split/2.Simon MacMullen2014-08-211-6/+7
| | |
| * | Merge bug25214Simon MacMullen2014-08-213-17/+35
| |\ \
| | * | Rename boolean to reuse / single.Simon MacMullen2014-08-213-17/+17
| | | |
| | * | Make Mnesia tx worker pool jobs use a disposable process so that if ↵Simon MacMullen2014-08-213-17/+35
| |/ / | | | | | | | | | mnesia_locker decides to randomly send a message there later it will just get dropped and not cause chaos.
* | | Unmislead comment.Simon MacMullen2014-08-211-1/+1
| | |
* | | Don't leak supervisors.Simon MacMullen2014-08-211-1/+1
| | |
* | | Remove silly race.Simon MacMullen2014-08-211-5/+12
| | |
* | | Merge in defaultSimon MacMullen2014-08-205-15/+185
|\ \ \ | | |/ | |/|
| * | Merge bug26353Simon MacMullen2014-08-200-0/+0
| |\ \
| * | | Those diagnostics are actually a bit dubious, they insert quite a delay and ↵Simon MacMullen2014-08-201-2/+1
| |/ / | | | | | | | | | aren't really designed to be logged.
| * | call/2 -> call/3.Simon MacMullen2014-08-201-1/+2
| | |
| * | MErge bug25813Simon MacMullen2014-08-2010-84/+111
| |\ \
| | * \ stable to defaultSimon MacMullen2014-08-2055-833/+2145
| | |\ \ | | | |/ | | |/|
| | * | Merge bug26354Simon MacMullen2014-08-201-1/+1
| | |\ \
| | | * | Get the message format right for this case too.Simon MacMullen2014-08-201-1/+1
| | |/ /
| * | | Fix a race.Simon MacMullen2014-08-201-1/+7
| | | |
| * | | Return 1 consumer.Simon MacMullen2014-08-181-2/+3
| | | |
| * | | Implement declaration.Simon MacMullen2014-08-181-2/+40
| | | |
| * | | Prevent malicious channels from sending messages to arbitrary processes with ↵Simon MacMullen2014-08-182-3/+26
| | | | | | | | | | | | | | | | this mechanism.
| * | | We'll just document that mandatory doesn't work.Simon MacMullen2014-08-181-1/+0
| | | |
| * | | base64 encoding is expensive enough to make it worth precalculating the ↵Simon MacMullen2014-08-151-12/+14
| | | | | | | | | | | | | | | | encoded pid just in case we want to reuse it.
| * | | Remove support for decorators to modify routing for the default exchange; it ↵Simon MacMullen2014-08-141-6/+5
| | | | | | | | | | | | | | | | doesn't make a ton of sense. Also no longer describe this as an optimisation since some of it is more than that.
| * | | We don't need to worry about confirms, they will get sent. Change to a soft ↵Simon MacMullen2014-08-141-4/+6
| | | | | | | | | | | | | | | | error for new errors.
| * | | Support cancel.Simon MacMullen2014-08-141-1/+9
| | | |
| * | | Unused var.Simon MacMullen2014-08-141-1/+1
| | | |
| * | | Security hole--Simon MacMullen2014-08-142-21/+25
| | | |
| * | | Rename this thing for consistency and also RabbitMQ-specificness.Simon MacMullen2014-08-142-6/+7
| | | |
| * | | Prototype of fast RPC replies.Simon MacMullen2014-08-132-5/+92
| | | |
* | | | CosmeticSimon MacMullen2014-08-201-2/+1
| | | |
* | | | Make this test make sense for the new regime. There will of course need to ↵Simon MacMullen2014-08-201-8/+4
| | | | | | | | | | | | | | | | be lots of new tests but at least this passes now.
* | | | Switch to making our restart decisions based on explicitly determining ↵Simon MacMullen2014-08-204-96/+78
| | | | | | | | | | | | | | | | whether we are starting for the first time or not. This is not very OTPish but it turns out to be necessary: there's no way to distinguish between losing a race to declare, starting a new slave, and restarting to a new slave otherwise. As an upside this code is shorter and more obviously correct (to me at least).
* | | | Allow crashing queue processes to come back as slaves if that's what we need.Simon MacMullen2014-08-202-1/+10
| | | |
* | | | This is perhaps more honest; if the queue was transient we will not recover ↵Simon MacMullen2014-08-201-2/+1
| | | | | | | | | | | | | | | | anything and if the queue is mirrored we will restart a slave and then also not recover anything.
View Lorry Controller Status