Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Recover e2e properly. | Simon MacMullen | 2011-03-31 | 2 | -10/+12 |
| | |||||
* | Fix tests. | Simon MacMullen | 2011-03-31 | 1 | -1/+1 |
| | |||||
* | Unify recovery into one boot step, based binding recovery on the queues that ↵ | Simon MacMullen | 2011-03-31 | 4 | -62/+55 |
| | | | | have been recovered. | ||||
* | Oops. | Simon MacMullen | 2011-03-31 | 1 | -1/+1 |
| | |||||
* | Only recover bindings that are to exchanges or to queues that are on this node. | Simon MacMullen | 2011-03-31 | 2 | -9/+23 |
| | |||||
* | Change exchange type API to not distinguish between creating and recovering, ↵ | Simon MacMullen | 2011-03-31 | 9 | -48/+62 |
| | | | | and to allow recovering bindings. Recover bindings when needed. | ||||
* | Only recover exchanges and bindings that need to be recovered. | Simon MacMullen | 2011-03-30 | 2 | -7/+13 |
| | |||||
* | Make rabbitmq-server compilable. | Simon MacMullen | 2011-03-30 | 1 | -2/+2 |
| | |||||
* | merge bug23993 into default (OCF script is broken when specifying an ↵ | Alexandru Scvortov | 2011-03-30 | 4 | -146/+142 |
|\ | | | | | | | alternative config file) | ||||
| * | cosmetic: put 'rabbitmqctl wait' code in the right place | Matthias Radestock | 2011-03-30 | 1 | -3/+7 |
| | | |||||
| * | Shutup, dialyzer | Matthew Sackman | 2011-03-30 | 1 | -2/+7 |
| | | |||||
| * | cosmetic | Matthias Radestock | 2011-03-30 | 1 | -70/+60 |
| | | |||||
| * | test refactor: mock the writer only once | Matthias Radestock | 2011-03-30 | 1 | -33/+11 |
| | | |||||
| * | cosmetic | Matthias Radestock | 2011-03-29 | 1 | -17/+10 |
| | | |||||
| * | english | Matthew Sackman | 2011-03-29 | 1 | -5/+5 |
| | | |||||
| * | Improve documentation of BQ concerning the issuance of confirms | Matthew Sackman | 2011-03-29 | 1 | -7/+31 |
| | | |||||
| * | cosmetic | Matthias Radestock | 2011-03-28 | 1 | -5/+5 |
| | | |||||
| * | tweak: only invoke rabbit_exchange:callback when absolutely necessary | Matthias Radestock | 2011-03-28 | 1 | -1/+1 |
| | | |||||
| * | more sensible order of exchange exports | Matthias Radestock | 2011-03-28 | 1 | -8/+10 |
| | | |||||
* | | Correct test for existance of config filebug23993 | Matthew Sackman | 2011-03-25 | 1 | -4/+4 |
|/ | |||||
* | Merging bug23831 to default | Matthew Sackman | 2011-03-25 | 0 | -0/+0 |
|\ | |||||
* | | removing trap_exit flag in rabbit_channel | Vlad Alexandru Ionescu | 2011-03-23 | 1 | -1/+0 |
|/ | |||||
* | cough | Matthew Sackman | 2011-03-23 | 1 | -2/+2 |
| | |||||
* | Merging bug23425 to default | Matthew Sackman | 2011-03-23 | 8 | -151/+486 |
|\ | |||||
| * | Turns out it's very important that we do write the schema_version when call ↵bug23425 | Matthew Sackman | 2011-03-23 | 1 | -1/+2 |
| | | | | | | | | mnesia:cluster | ||||
| * | Abstract out continuation | Matthew Sackman | 2011-03-23 | 2 | -16/+15 |
| | | |||||
| * | Merge heads | Simon MacMullen | 2011-03-23 | 13 | -141/+185 |
| |\ | |||||
| | * | Detect discnodeishness prior to suffering disclessness | Matthew Sackman | 2011-03-23 | 1 | -1/+3 |
| | | | |||||
| | * | sort out how often and when we do the rabbit:prepare (set up log handlers), ↵ | Matthew Sackman | 2011-03-23 | 3 | -10/+7 |
| | | | | | | | | | | | | and actually make that do the mnesia upgrade. | ||||
| | * | New decree is that you're not meant to match against ensure_stuff calls | Matthew Sackman | 2011-03-23 | 2 | -7/+11 |
| | | | |||||
| | * | Merge default into bug23425 | Matthew Sackman | 2011-03-23 | 8 | -147/+485 |
| | |\ | |||||
| | | * | Renaming bits and pieces for consistency and checking a few more return ↵ | Matthew Sackman | 2011-03-22 | 2 | -19/+21 |
| | | | | | | | | | | | | | | | | values, plus other minor fixes | ||||
| * | | | Record all nodes, don't list them when we refuse to start. | Simon MacMullen | 2011-03-23 | 3 | -23/+17 |
| | |/ | |/| | |||||
| * | | Better name, vertical space. | Simon MacMullen | 2011-03-21 | 1 | -13/+11 |
| | | | |||||
| * | | When upgrading a secondary node we call init_db twice: once early to force a ↵ | Simon MacMullen | 2011-03-21 | 2 | -13/+18 |
| | | | | | | | | | | | | cluster rejoin (at which point we are not ready to do local upgrades, e.g. fhc is not running) and then once at the regular time. Deal with that. | ||||
| * | | Don't try to determine whether a backup is needed before doing anything, ↵ | Simon MacMullen | 2011-03-21 | 1 | -40/+18 |
| | | | | | | | | | | | | take it as needed. This inverts the backup and the lock file - the backup now comes first and the lock file is only used to defend apply_upgrades/3. | ||||
| * | | Oops. | Simon MacMullen | 2011-03-21 | 1 | -2/+2 |
| | | | |||||
| * | | maybe_remove_backup is safe when returning version_not_available since we ↵ | Simon MacMullen | 2011-03-21 | 1 | -10/+9 |
| | | | | | | | | | | | | would not have taken a backup in the first place. However, this is not exactly obvious, so let's not do that. | ||||
| * | | Merge heads | Simon MacMullen | 2011-03-21 | 1 | -22/+18 |
| |\ \ | |||||
| | * | | Various QA tidyups, and stop exporting the backup / remove backup functions. | Simon MacMullen | 2011-03-21 | 1 | -22/+18 |
| | | | | |||||
| * | | | Well I thought =~= was beautiful and appropriately approximate to == | Matthew Sackman | 2011-03-18 | 3 | -5/+5 |
| |/ / | |||||
| * | | Take a single backup before any upgrade, remove it when we're all clear. | Simon MacMullen | 2011-03-18 | 1 | -37/+69 |
| | | | |||||
| * | | Merge heads | Simon MacMullen | 2011-03-18 | 3 | -21/+20 |
| |\ \ | |||||
| | * | | Can't call =~= with non-version args, thus shuffle things around a bit. End ↵ | Matthew Sackman | 2011-03-17 | 1 | -17/+16 |
| | | | | | | | | | | | | | | | | up saving a line. | ||||
| | * | | Improve symmetry: if we write the running_disc_nodes on rabbit shutdown, we ↵ | Matthew Sackman | 2011-03-17 | 2 | -2/+2 |
| | | | | | | | | | | | | | | | | should nuke it on rabbit startup. This then means that the prelaunch thingy is always run with the previously_running_disc_nodes file present. I believe this makes no semantic changes, but the improved symmetry is worth having | ||||
| | * | | rabbit:stop() is not always called when rabbit is stopping. E.g. q(). ↵ | Matthew Sackman | 2011-03-17 | 2 | -2/+2 |
| | | | | | | | | | | | | | | | | doesn't invoke rabbit:stop/0. It does invoke rabbit:stop/1 though. | ||||
| * | | | Ensure mnesia is stopped for the local upgrade backup. | Simon MacMullen | 2011-03-18 | 1 | -2/+3 |
| |/ / | |||||
| * | | If we have version_not_available, then it makes sense to have ↵ | Matthew Sackman | 2011-03-17 | 1 | -2/+1 |
| | | | | | | | | | | | | version_mismatch, not schema_mismatch | ||||
| * | | renamings | Matthew Sackman | 2011-03-17 | 3 | -35/+34 |
| | | | |||||
| * | | incorporate qa feedback. The version.erl api is rather nice now: the version ↵ | Matthew Sackman | 2011-03-17 | 3 | -146/+139 |
| | | | | | | | | | | | | itself is entirely opaque - whilst it can be read, there's nothing provided to decompose it at all. |