Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

queue stats logging refinements #1875

Open
wants to merge 7 commits into
base: develop-2.9
Choose a base branch
from
Open

queue stats logging refinements #1875

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
b219714
suppress logging at info level when queue stats are 0 in reaper and e…
hmmr Nov 1, 2023
ccdd0e8
expose replrtq_logfrequency tunable
hmmr Nov 1, 2023
c3f2851
do replrtq stats logging with metadata with queue_name
hmmr Nov 1, 2023
9641ace
suppress logging of stats with empty queue sizes
hmmr Nov 1, 2023
f43c0f1
better name for tunable controlling queue stats logging frequency
hmmr Nov 4, 2023
c1f9af8
redo 9641acea, now via a new tunable, queue_manager_log_suppress_zero…
hmmr Nov 4, 2023
afa0040
fixup for 9641acea10df
hmmr Nov 5, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 14 additions & 1 deletion priv/riak_kv.schema
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1146,6 +1146,19 @@
{default, 24}
]}.

%% @doc Periodically log replrtq queue stats (queue size, reap/delete
%% attempts and aborts) at this interval (seconds)
{mapping, "queue_manager_log_frequency", "riak_kv.queue_manager_log_frequency", [
{datatype, integer},
{default, 30}
]}.

%% @doc Suppress logging of queue stats when all items are 0
{mapping, "queue_manager_log_suppress_zero_stats", "riak_kv.queue_manager_log_suppress_zero_stats", [
{datatype, flag},
{default, off}
]}.

%% @doc The maximume number of workers to be for any given peer may be
%% configured - if not configured will default to the number of sinkworkers
{mapping, "replrtq_sinkpeerlimit", "riak_kv.replrtq_sinkpeerlimit", [
Expand All @@ -1171,4 +1184,4 @@
{datatype, {flag, enabled, disabled}},
{default, disabled},
{commented, enabled}
]}.
]}.
19 changes: 13 additions & 6 deletions src/riak_kv_eraser.erl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -292,12 +292,19 @@ handle_async_message(timeout, State) ->
pqueue_length = {RedoQL, DeleteQL - BatchSize}}}
end;
handle_async_message(log_queue, State) ->
lager:info("Eraser job ~w has queue lengths ~w " ++
"delete_attempts=~w delete_aborts=~w ",
[State#state.job_id,
State#state.pqueue_length,
State#state.delete_attempts,
State#state.delete_aborts]),
case app_helper:get_env(riak_kv, queue_manager_log_suppress_zero_stats, false) of
true when State#state.pqueue_length =:= {0,0},
State#state.delete_attempts =:= 0,
State#state.delete_aborts =:= 0 ->
nop;
_ ->
lager:info("Reaper Job ~w has queue lengths ~w "
"reap_attempts=~w reap_aborts=~w",
[State#state.job_id,
State#state.pqueue_length,
State#state.delete_attempts,
State#state.delete_aborts])
end,
erlang:send_after(?LOG_TICK, self(), log_queue),
{noreply, State#state{delete_attempts = 0, delete_aborts = 0}}.

Expand Down
19 changes: 13 additions & 6 deletions src/riak_kv_reaper.erl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -275,12 +275,19 @@ handle_async_message(timeout, State) ->
pqueue_length = {RedoQL, ReapQL - BatchSize}}}
end;
handle_async_message(log_queue, State) ->
lager:info("Reaper Job ~w has queue lengths ~w " ++
" reap_attempts=~w reap_aborts=~w ",
[State#state.job_id,
State#state.pqueue_length,
State#state.reap_attempts,
State#state.reap_aborts]),
case app_helper:get_env(riak_kv, queue_manager_log_suppress_zero_stats, false) of
true when State#state.pqueue_length =:= {0,0},
State#state.reap_attempts =:= 0,
State#state.reap_aborts =:= 0 ->
nop;
_ ->
lager:info("Reaper Job ~w has queue lengths ~w "
"reap_attempts=~w reap_aborts=~w",
[State#state.job_id,
State#state.pqueue_length,
State#state.reap_attempts,
State#state.reap_aborts])
end,
erlang:send_after(?LOG_TICK, self(), log_queue),
{noreply, State#state{reap_attempts = 0, reap_aborts = 0}}.

Expand Down
14 changes: 10 additions & 4 deletions src/riak_kv_replrtq_snk.erl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -393,10 +393,14 @@ handle_cast({requeue_work, WorkItem}, State) ->

handle_info(timeout, State) ->
prompt_work(),
erlang:send_after(?LOG_TIMER_SECONDS * 1000, self(), log_stats),
LogFreq = app_helper:get_env(riak_kv, queue_manager_log_frequency,
?LOG_TIMER_SECONDS),
erlang:send_after(LogFreq * 1000, self(), log_stats),
{noreply, State};
handle_info(log_stats, State) ->
erlang:send_after(?LOG_TIMER_SECONDS * 1000, self(), log_stats),
LogFreq = app_helper:get_env(riak_kv, queue_manager_log_frequency,
?LOG_TIMER_SECONDS),
erlang:send_after(LogFreq * 1000, self(), log_stats),
SinkWork0 =
case State#state.enabled of
true ->
Expand Down Expand Up @@ -782,7 +786,8 @@ log_mapfun({QueueName, Iteration, SinkWork}) ->
{replmod_time, RT},
{modified_time, MTS, MTM, MTH, MTD, MTL}}
= SinkWork#sink_work.queue_stats,
lager:info("Queue=~w success_count=~w error_count=~w" ++
lager:info([{queue_name, QueueName}],
"Queue=~w success_count=~w error_count=~w" ++
" mean_fetchtime_ms=~s" ++
" mean_pushtime_ms=~s" ++
" mean_repltime_ms=~s" ++
Expand All @@ -796,7 +801,8 @@ log_mapfun({QueueName, Iteration, SinkWork}) ->
end,
PeerDelays =
lists:foldl(FoldPeerInfoFun, "", SinkWork#sink_work.peer_list),
lager:info("Queue=~w has peer delays of~s", [QueueName, PeerDelays]),
lager:info([{queue_name, QueueName}],
"Queue=~w has peer delays of~s", [QueueName, PeerDelays]),
{QueueName, Iteration, SinkWork#sink_work{queue_stats = ?ZERO_STATS}}.


Expand Down
15 changes: 10 additions & 5 deletions src/riak_kv_replrtq_src.erl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -288,14 +288,14 @@ init([]) ->
QC = lists:map(MaptoQC, QFM),
QL = app_helper:get_env(riak_kv, replrtq_srcqueuelimit, ?QUEUE_LIMIT),
OL = app_helper:get_env(riak_kv, replrtq_srcobjectlimit, ?OBJECT_LIMIT),
LogFreq = app_helper:get_env(riak_kv, replrtq_logfrequency, ?LOG_TIMER_SECONDS * 1000),
LogFreq = app_helper:get_env(riak_kv, queue_manager_log_frequency, ?LOG_TIMER_SECONDS),
erlang:send_after(LogFreq, self(), log_queue),
{ok, #state{queue_filtermap = QFM,
queue_map = QM,
queue_countmap = QC,
queue_limit = QL,
object_limit = OL,
log_frequency_in_ms = LogFreq}}.
log_frequency_in_ms = LogFreq * 1000}}.

handle_call({rtq_ttaaefs, QueueName, ReplEntries}, _From, State) ->
{ApproachingLimit, QueueMap, QueueCountMap} =
Expand Down Expand Up @@ -408,10 +408,15 @@ handle_cast({rtq_coordput, Bucket, ReplEntry}, State) ->
queue_countmap = QueueCountMap}}.

handle_info(log_queue, State) ->
Suppress = app_helper:get_env(riak_kv, queue_manager_log_suppress_zero_stats, false),
LogFun =
fun({QueueName, {P1Q, P2Q, P3Q}}) ->
lager:info("QueueName=~w has queue sizes p1=~w p2=~w p3=~w",
[QueueName, P1Q, P2Q, P3Q])
fun({_QueueName, {0, 0, 0}}) when Suppress == true ->
ok;
({QueueName, {P1Q, P2Q, P3Q}}) ->
lager:info(
[{queue_name, QueueName}, {p1q, P1Q}, {p2q, P2Q}, {p3q, P3Q}],
"QueueName=~w has queue sizes p1=~w p2=~w p3=~w",
[QueueName, P1Q, P2Q, P3Q])
end,
lists:foreach(LogFun, State#state.queue_countmap),
erlang:send_after(State#state.log_frequency_in_ms, self(), log_queue),
Expand Down