25
1
mirror of https://github.com/processone/ejabberd.git synced 2024-12-22 17:28:25 +01:00

Add new options for OOM watchdog

* oom_watermark: 1..100
  Start OOM watchdog only when system memory usage exceeds
  this value in percents. When the usage drops below the value,
  OOM watchdog is stopped. The default is 80 (percents).
  Note that once OOM watchdog is started, it performs full garbage
  collection periodically: this can be seen as spikes in CPU
  utilization and drops in RAM usage. If your system is permanently
  above the watermark, it may cause significant CPU overhead.

* oom_queue: positive integer
  Only trigger OOM killer when total amount of messages in all queues
  of all Erlang processes is above this value. The default is 10000.
  Note that this value only takes effect when `oom_killer` is set
  to `true` (this is the default). Otherwise, only a warning will
  be logged.
This commit is contained in:
Evgeniy Khramtsov 2018-06-21 14:35:19 +03:00
parent 274e9fe7b5
commit 55f8aa1b22
2 changed files with 22 additions and 8 deletions

View File

@ -50,9 +50,9 @@ start(normal, _Args) ->
ejabberd_mnesia:start(),
file_queue_init(),
maybe_add_nameservers(),
ejabberd_system_monitor:start(),
case ejabberd_sup:start_link() of
{ok, SupPid} ->
ejabberd_system_monitor:start(),
register_elixir_config_hooks(),
ejabberd_cluster:wait_for_sync(infinity),
{T2, _} = statistics(wall_clock),

View File

@ -31,7 +31,7 @@
-author('ekhramtsov@process-one.net').
%% API
-export([start/0, opt_type/1]).
-export([start/0, opt_type/1, config_reloaded/0]).
%% gen_event callbacks
-export([init/1, handle_event/2, handle_call/2,
@ -67,15 +67,21 @@ start() ->
application:set_env(os_mon, start_os_sup, false),
application:set_env(os_mon, start_memsup, true),
application:set_env(os_mon, start_disksup, false),
ejabberd:start_app(os_mon).
ejabberd:start_app(os_mon),
set_oom_watermark().
excluded_apps() ->
[os_mon, mnesia, sasl, stdlib, kernel].
-spec config_reloaded() -> ok.
config_reloaded() ->
set_oom_watermark().
%%%===================================================================
%%% gen_event callbacks
%%%===================================================================
init([]) ->
ejabberd_hooks:add(config_reloaded, ?MODULE, config_reloaded, 50),
{ok, #state{}}.
handle_event({set_alarm, {system_memory_high_watermark, _}}, State) ->
@ -112,7 +118,7 @@ handle_info(Info, State) ->
{ok, State}.
terminate(_Reason, _State) ->
ok.
ejabberd_hooks:delete(config_reloaded, ?MODULE, config_reloaded, 50).
code_change(_OldVsn, State, _Extra) ->
{ok, State}.
@ -128,7 +134,8 @@ handle_overload(State) ->
handle_overload(_State, Procs) ->
AppPids = get_app_pids(),
{TotalMsgs, ProcsNum, Apps, Stats} = overloaded_procs(AppPids, Procs),
if TotalMsgs >= 10000 ->
MaxMsgs = ejabberd_config:get_option(oom_queue, 10000),
if TotalMsgs >= MaxMsgs ->
SortedStats = lists:reverse(lists:keysort(#proc_stat.qlen, Stats)),
error_logger:warning_msg(
"The system is overloaded with ~b messages "
@ -312,14 +319,21 @@ kill_proc(Pid) ->
exit(Pid, kill),
Pid.
-spec set_oom_watermark() -> ok.
set_oom_watermark() ->
WaterMark = ejabberd_config:get_option(oom_watermark, 80),
memsup:set_sysmem_high_watermark(WaterMark/100).
-spec maybe_restart_app(atom()) -> any().
maybe_restart_app(lager) ->
ejabberd_logger:restart();
maybe_restart_app(_) ->
ok.
-spec opt_type(oom_killer) -> fun((boolean()) -> boolean());
(atom()) -> [atom()].
opt_type(oom_killer) ->
fun(B) when is_boolean(B) -> B end;
opt_type(_) -> [oom_killer].
opt_type(oom_watermark) ->
fun(I) when is_integer(I), I>0, I<100 -> I end;
opt_type(oom_queue) ->
fun(I) when is_integer(I)>0 -> I end;
opt_type(_) -> [oom_killer, oom_watermark, oom_queue].