Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Nicer feedback.bug26475 | Simon MacMullen | 2014-12-09 | 1 | -1/+3 |
| | |||||
* | Split out preparation from rename/2, and add comments. | Simon MacMullen | 2014-12-09 | 1 | -29/+77 |
| | |||||
* | Various fixes / improvements: with the new scheme multiple renamings will ↵ | Simon MacMullen | 2014-12-09 | 1 | -32/+39 |
| | | | | not waork, so check for them. Fix abortive renames (with a copy of the mnesia dir). Ensure we don't talk to any other nodes while backing up / restoring since that inevitably goes wrong. Split up rename/2 a bit and rename a few other things for more clarity. | ||||
* | Fail clearly if the node specified with '-n' is not a member of the cluster ↵ | Simon MacMullen | 2014-12-09 | 1 | -4/+6 |
| | | | | before or after. | ||||
* | Ensure become(node()) works rather than failing with {node_running, ...}. | Simon MacMullen | 2014-12-09 | 1 | -2/+2 |
| | |||||
* | WIP clustererer compatibility: install the fallback and "upgrade" mnesia at ↵ | Simon MacMullen | 2014-12-08 | 1 | -13/+14 |
| | | | | rename_cluster_node time, instead of during next boot. This means we need to faff around a bit making sure we do not contact other nodes during the upgrade, but also that on-disk data structures are left in a consistent state for the clusterererer to find them. | ||||
* | Maybe clearer? | Simon MacMullen | 2014-12-01 | 1 | -1/+1 |
| | |||||
* | Add a section to the manual page. | Simon MacMullen | 2014-12-01 | 1 | -0/+35 |
| | |||||
* | Swap location of backup and regular config files. A bit less logical, but we ↵ | Simon MacMullen | 2014-12-01 | 1 | -4/+4 |
| | | | | need them to be in the right place when starting for big-bang upgrades to work. | ||||
* | Put all the rename-related temp files in a directory together, and keep ↵ | Simon MacMullen | 2014-12-01 | 1 | -13/+21 |
| | | | | copies of the rewritten configs so we can roll back without having to re-edit them. | ||||
* | Allow argument-pairs to rename_cluster_node to go in any order (and thus ↵ | Simon MacMullen | 2014-12-01 | 2 | -23/+46 |
| | | | | allow remote-only renamings). Don't use mnesia:transform_table/3 since it's rather picky about which disk nodes are up. | ||||
* | Rename again | Simon MacMullen | 2014-11-28 | 1 | -3/+4 |
| | |||||
* | And unbreak clusters. | Simon MacMullen | 2014-11-28 | 1 | -0/+1 |
| | |||||
* | Detect starting as the old node name after a rename and roll back. Also ↵ | Simon MacMullen | 2014-11-28 | 2 | -26/+44 |
| | | | | rename a bit. | ||||
* | Less io:format, more rabbit_log. | Simon MacMullen | 2014-11-28 | 1 | -7/+6 |
| | |||||
* | Improve the chance of not losing a bunch of data. | Simon MacMullen | 2014-11-28 | 1 | -0/+1 |
| | |||||
* | Oops | Simon MacMullen | 2014-11-28 | 1 | -1/+1 |
| | |||||
* | Rename things | Simon MacMullen | 2014-11-28 | 2 | -9/+8 |
| | |||||
* | Remove commented out bit, remove pointless RPC (we're in the cluster by that ↵ | Simon MacMullen | 2014-11-28 | 1 | -47/+13 |
| | | | | point, we can make Mnesia changes ourselves). | ||||
* | We are presumably the source of truth now, so make sure we boot. | Simon MacMullen | 2014-11-28 | 1 | -0/+1 |
| | |||||
* | Support renaming other remote nodes at the same time, in case we want to do ↵ | Simon MacMullen | 2014-11-27 | 2 | -68/+69 |
| | | | | a big bang. | ||||
* | Don't be so keep to call upgrade_mode/1, it can die in all sorts of ways. | Simon MacMullen | 2014-11-27 | 2 | -30/+35 |
| | |||||
* | Rearrange things a bit. We now have something that seems to work. | Simon MacMullen | 2014-11-27 | 3 | -21/+55 |
| | | | | | | | | | | | | | | | | | | | Merging a modified database back into a running cluster seems to be a bit of a non-starter; Mnesia blows up with "Failed to merge schema: Incompatible schema cookies." So instead we do something similar to if we were doing upgrades: we always create the modified backup, but only load it if we are the first node in the cluster to start; if not we reset our mnesia state, resync with the cluster, then tell the cluster to forget our old incarnation and update its queue records for the new one. This also has the advantage that we don't need two rename commands, we just have the local offline one. Which is hopefully nicer operationally. It does mean that we go to the trouble of taking the backup and modifying it only to throw it away if we turn out to be a secondary node, but it should not be too expensive. | ||||
* | WIP towards rabbitmqctl rename_other_node. | Simon MacMullen | 2014-11-26 | 2 | -1/+29 |
| | |||||
* | Oops | Simon MacMullen | 2014-11-26 | 1 | -1/+1 |
| | |||||
* | Create a way to mutate pids and use that rather than creating "fake" pids. | Simon MacMullen | 2014-11-26 | 3 | -11/+24 |
| | |||||
* | First pass at "rabbitmqctl rename_current_node" | Simon MacMullen | 2014-11-26 | 2 | -3/+139 |
| | |||||
* | stable to default | Simon MacMullen | 2014-11-26 | 17 | -187/+507 |
|\ | |||||
| * | Merge bug26471 | Simon MacMullen | 2014-11-26 | 16 | -178/+493 |
| |\ | |||||
| | * | Removed tag rabbitmq_v3_4_2 | Simon MacMullen | 2014-11-26 | 0 | -0/+0 |
| | | | |||||
| | * | Added tag rabbitmq_v3_4_2 for changeset d472925bb786 | Simon MacMullen | 2014-11-26 | 0 | -0/+0 |
| | | | |||||
| | * | stable to default | Simon MacMullen | 2014-11-25 | 1 | -0/+6 |
| | |\ | |||||
| | * \ | stable to default | Simon MacMullen | 2014-11-25 | 1 | -7/+29 |
| | |\ \ | |||||
| | * \ \ | Merge heads of default | Jean-Sebastien Pedron | 2014-11-24 | 1 | -4/+4 |
| | |\ \ \ | |||||
| | | * \ \ | Merge bug26257 (again) | Jean-Sebastien Pedron | 2014-11-24 | 4 | -7/+23 |
| | | |\ \ \ | |||||
| | | * | | | | When reporting invalid cluster node names, keep configuration orderbug26257 | Jean-Sebastien Pedron | 2014-11-24 | 1 | -4/+4 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | While here, fix an unused variable warning. | ||||
| | * | | | | | stable to default | Simon MacMullen | 2014-11-24 | 16 | -178/+493 |
| | |\ \ \ \ \ | | | | |/ / / | | | |/| | | | |||||
| | | * | | | | Merge bug26257 | Simon MacMullen | 2014-11-21 | 1 | -14/+34 |
| | | |\ \ \ \ | | | | |/ / / | |||||
| | | | * | | | Detect other invalid 'cluster_nodes' values | Jean-Sebastien Pedron | 2014-11-21 | 1 | -7/+33 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | In all cases, abort startup with an explanatory message. | ||||
| | | | * | | | Remove support for the legacy 'cluster_nodes' values | Jean-S?bastien P?dron | 2014-11-20 | 1 | -13/+7 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Before this change, a list of nodes without the node type was accepted. In this case, the node type was guessed and a warning suggesting how to update the configuration was logged. Now, the node type is mandatory and the RabbitMQ server refuses to start if the node type is unspecified. | ||||
| | | * | | | | stable to default | Simon MacMullen | 2014-11-20 | 3 | -15/+13 |
| | | |\ \ \ \ | |||||
| | | * \ \ \ \ | stable to default | Simon MacMullen | 2014-11-20 | 1 | -1/+13 |
| | | |\ \ \ \ \ | |||||
| | | * \ \ \ \ \ | stable to default | Simon MacMullen | 2014-11-20 | 1 | -1/+7 |
| | | |\ \ \ \ \ \ | | | | |_|_|/ / / | | | |/| | | | | | |||||
| | | * | | | | | | stable to default | Simon MacMullen | 2014-11-19 | 1 | -0/+5 |
| | | |\ \ \ \ \ \ | |||||
| | | * \ \ \ \ \ \ | Merge bug26469 | Simon MacMullen | 2014-11-14 | 11 | -96/+201 |
| | | |\ \ \ \ \ \ \ | |||||
| | | | * | | | | | | | Tweak the APIs again, so that authz plugins aren't expected to create a ↵bug26469 | Simon MacMullen | 2014-11-14 | 8 | -71/+120 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | #auth_user record in the first place, just whatever impl they want. Which necessitates seperate login functions for authz and authn, and if we're going to do that we might as well split the behaviours so that we have the possibility of making an authz-only plugin. | ||||
| | | | * | | | | | | | Rename. | Simon MacMullen | 2014-11-14 | 1 | -2/+2 |
| | | | | | | | | | | | |||||
| | | | * | | | | | | | Fix tests. | Simon MacMullen | 2014-11-14 | 1 | -4/+6 |
| | | | | | | | | | | | |||||
| | | | * | | | | | | | Never pass the #user{} record into auth backends, they should only see their ↵ | Simon MacMullen | 2014-11-13 | 7 | -62/+87 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | own stuff. Get rid of authN_backend, it has little reason for existing. Flatten case of authZ_backend. | ||||
| | | | * | | | | | | | Cosmetic. | Simon MacMullen | 2014-11-13 | 1 | -31/+30 |
| | | | | | | | | | | |