diff options
| author | Simon MacMullen <simon@rabbitmq.com> | 2014-10-09 16:09:48 +0100 |
|---|---|---|
| committer | Simon MacMullen <simon@rabbitmq.com> | 2014-10-09 16:09:48 +0100 |
| commit | 29d307e9a9c2632f40f14dba35ff244dffaeaa46 (patch) | |
| tree | e6424c13c8cbf32ab92a37eb589831798255a3a0 | |
| parent | 502e7335316508845080722b063339e9d59a2a9f (diff) | |
| download | rabbitmq-server-29d307e9a9c2632f40f14dba35ff244dffaeaa46.tar.gz | |
Slightly more accurate comment.
| -rw-r--r-- | src/rabbit_node_monitor.erl | 9 |
1 files changed, 5 insertions, 4 deletions
diff --git a/src/rabbit_node_monitor.erl b/src/rabbit_node_monitor.erl index fdda6252..933beed7 100644 --- a/src/rabbit_node_monitor.erl +++ b/src/rabbit_node_monitor.erl @@ -622,10 +622,11 @@ cast(Node, Msg) -> gen_server:cast({?SERVER, Node}, Msg). %%-------------------------------------------------------------------- %% mnesia:system_info(db_nodes) (and hence -%% rabbit_mnesia:cluster_nodes(running)) does not give reliable -%% results when partitioned. So we have a small set of replacement -%% functions here. "rabbit" in a function's name implies we test if -%% the rabbit application is up, not just the node. +%% rabbit_mnesia:cluster_nodes(running)) does not return all nodes +%% when partitioned, just those that we are sharing Mnesia state +%% with. So we have a small set of replacement functions +%% here. "rabbit" in a function's name implies we test if the rabbit +%% application is up, not just the node. %% As we use these functions to decide what to do in pause_minority %% state, they *must* be fast, even in the case where TCP connections |