File tree Expand file tree Collapse file tree 1 file changed +8
-6
lines changed Expand file tree Collapse file tree 1 file changed +8
-6
lines changed Original file line number Diff line number Diff line change @@ -72,12 +72,14 @@ defmodule Phoenix.Tracker do
72
72
73
73
## Application Shutdown
74
74
75
- A tracker does not automatically replicate its state across the cluster as it
76
- shuts down. This means that your supervision tree shuts down normally - as it
77
- does when you call `System.stop()` or the BEAM receives a `SIGTERM` - any
78
- presences that the local tracker instance has will continue to be seen as
79
- present by other trackers in the cluster until the `:down_period` for the
80
- instance has passed.
75
+ When a tracker shuts down, the other nodes do not assume it has been gone
76
+ for good. After all, in a distributed system, it is impossible to know if something
77
+ is just temporarily unavailable or if it has crashed.
78
+
79
+ For this reason, when you call `System.stop()` or the Erlang VM receives a
80
+ `SIGTERM` - any presences that the local tracker instance has will continue to
81
+ be seen as present by other trackers in the cluster until the `:down_period`
82
+ for the instance has passed.
81
83
82
84
If you want a normal shutdown to immediately cause other nodes to see that
83
85
tracker's presences as leaving, pass `permdown_on_shutdown: true`. On the
You can’t perform that action at this time.
0 commit comments