fix(ds): Apply review remarks

This commit is contained in:
ieQu1 2024-05-21 22:07:03 +02:00
parent 5c78ecba40
commit 59a09fb86f
No known key found for this signature in database
GPG Key ID: 488654DF3FED6FDE
3 changed files with 26 additions and 3 deletions

View File

@ -115,11 +115,11 @@ get_servers_local_preferred(DB, Shard) ->
Servers when is_list(Servers) ->
ok
end,
case lists:keyfind(node(), 2, Servers) of
case lists:keytake(node(), 2, Servers) of
false ->
Servers;
Local when is_tuple(Local) ->
[Local | lists:delete(Local, Servers)]
{value, Local, Rest} ->
[Local | Rest]
end.
lookup_leader(DB, Shard) ->

View File

@ -553,6 +553,17 @@ delete_next_until(
end.
handle_event(_ShardId, State = #s{gvars = Gvars}, Time, tick) ->
%% If the last message was published more than one epoch ago, and
%% the shard remains idle, we need to advance safety cutoff
%% interval to make sure the last epoch becomes visible to the
%% readers.
%%
%% We do so by emitting a dummy event that will be persisted by
%% the replication layer. Processing it will advance the
%% replication layer's clock.
%%
%% This operation is latched to avoid publishing events on every
%% tick.
case ets:lookup(Gvars, ?IDLE_DETECT) of
[{?IDLE_DETECT, Latch, LastWrittenTs}] ->
ok;
@ -562,6 +573,8 @@ handle_event(_ShardId, State = #s{gvars = Gvars}, Time, tick) ->
end,
case Latch of
false when ?EPOCH(State, Time) > ?EPOCH(State, LastWrittenTs) + 1 ->
%% Note: + 1 above delays the event by one epoch to add a
%% safety margin.
ets:insert(Gvars, {?IDLE_DETECT, true, LastWrittenTs}),
[dummy_event];
_ ->

View File

@ -0,0 +1,10 @@
Various fixes related to the `durable_sessions` feature:
- Add an option to execute read operations on the leader.
- `drop_generation` operation can be replayed multiple times by the replication layer, but it's not idempotent. This PR adds a workaround that avoids a crash when `drop_generation` doesn't succeed. In the future, however, we want to make `drop_generation` idempotent in a nicer way.
- Wrap storage layer events in a small structure containing the generation ID, to make sure events are handled by the same layout CBM & context that produced them.
- Fix crash when storage event arrives to the dropped generation (now removed `storage_layer:generation_at` function didn't handle the case of dropped generations).
- Implement `format_status` callback for several workers to minimize log spam
- Move the responsibility of `end_of_stream` detection to the layout CBM. Previously storage layer used a heuristic: old generations that return an empty batch won't produce more data. This was, obviously, incorrect: for example, bitfield-LTS layout MAY return empty batch while waiting for safe cutoff time.
- `reference` layout has been enabled in prod build. It could be useful for integration testing.
- Fix incorrect epoch calculation in `bitfield_lts:handle_event` callback that lead to missed safe cutoff time updates, and effectively, subscribers being unable to fetch messages until a fresh batch was published.