Skip to content

improve node deletion in hive (#7218) #10393

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

Merged
merged 1 commit into from
Oct 14, 2024
Merged

improve node deletion in hive (#7218) #10393

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
10 changes: 8 additions & 2 deletions ydb/core/mind/hive/hive_impl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
#include <ydb/core/cms/console/console.h>
#include <ydb/core/cms/console/configs_dispatcher.h>
#include <ydb/core/protos/counters_hive.pb.h>
#include <ydb/core/protos/node_broker.pb.h>
#include <ydb/core/util/tuples.h>
#include <ydb/library/yverify_stream/yverify_stream.h>
#include <ydb/library/actors/interconnect/interconnect.h>
Expand Down Expand Up @@ -96,10 +97,12 @@ void THive::RestartPipeTx(ui64 tabletId) {

bool THive::TryToDeleteNode(TNodeInfo* node) {
if (node->CanBeDeleted()) {
BLOG_I("TryToDeleteNode(" << node->Id << "): deleting");
DeleteNode(node->Id);
return true;
}
if (!node->DeletionScheduled) {
BLOG_D("TryToDeleteNode(" << node->Id << "): waiting " << GetNodeDeletePeriod());
Schedule(GetNodeDeletePeriod(), new TEvPrivate::TEvDeleteNode(node->Id));
node->DeletionScheduled = true;
}
Expand Down Expand Up @@ -987,8 +990,9 @@ void THive::OnActivateExecutor(const TActorContext&) {
BuildLocalConfig();
ClusterConfig = AppData()->HiveConfig;
SpreadNeighbours = ClusterConfig.GetSpreadNeighbours();
NodeBrokerEpoch = TDuration::MicroSeconds(NKikimrNodeBroker::TConfig().GetEpochDuration());
Send(NConsole::MakeConfigsDispatcherID(SelfId().NodeId()),
new NConsole::TEvConfigsDispatcher::TEvSetConfigSubscriptionRequest(NKikimrConsole::TConfigItem::HiveConfigItem));
new NConsole::TEvConfigsDispatcher::TEvSetConfigSubscriptionRequest({NKikimrConsole::TConfigItem::HiveConfigItem, NKikimrConsole::TConfigItem::NodeBrokerConfigItem}));
Execute(CreateInitScheme());
if (!ResponsivenessPinger) {
ResponsivenessPinger = new TTabletResponsivenessPinger(TabletCounters->Simple()[NHive::COUNTER_RESPONSE_TIME_USEC], TDuration::Seconds(1));
Expand Down Expand Up @@ -2208,7 +2212,9 @@ void THive::Handle(TEvHive::TEvInitiateTabletExternalBoot::TPtr& ev) {
void THive::Handle(NConsole::TEvConsole::TEvConfigNotificationRequest::TPtr& ev) {
const NKikimrConsole::TConfigNotificationRequest& record = ev->Get()->Record;
ClusterConfig = record.GetConfig().GetHiveConfig();
BLOG_D("Received TEvConsole::TEvConfigNotificationRequest with update of cluster config: " << ClusterConfig.ShortDebugString());
NodeBrokerEpoch = TDuration::MicroSeconds(record.GetConfig().GetNodeBrokerConfig().GetEpochDuration());
BLOG_D("Received TEvConsole::TEvConfigNotificationRequest with update of cluster config: " << ClusterConfig.ShortDebugString()
<< "; " << record.GetConfig().GetNodeBrokerConfig().ShortDebugString());
BuildCurrentConfig();
Send(ev->Sender, new NConsole::TEvConsole::TEvConfigNotificationResponse(record), 0, ev->Cookie);
}
Expand Down
7 changes: 6 additions & 1 deletion ydb/core/mind/hive/hive_impl.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -445,6 +445,7 @@ class THive : public TActor<THive>, public TTabletExecutedFlat, public THiveShar

NKikimrConfig::THiveConfig ClusterConfig;
NKikimrConfig::THiveConfig DatabaseConfig;
TDuration NodeBrokerEpoch;
std::unordered_map<TTabletTypes::EType, NKikimrConfig::THiveTabletLimit> TabletLimit; // built from CurrentConfig
std::unordered_map<TTabletTypes::EType, NKikimrHive::TDataCentersPreference> DefaultDataCentersPreference;
std::unordered_map<TDataCenterId, std::unordered_set<TNodeId>> RegisteredDataCenterNodes;
Expand Down Expand Up @@ -739,7 +740,11 @@ TTabletInfo* FindTabletEvenInDeleting(TTabletId tabletId, TFollowerId followerId
}

TDuration GetNodeDeletePeriod() const {
return TDuration::Seconds(CurrentConfig.GetNodeDeletePeriod());
if (CurrentConfig.HasNodeDeletePeriod()) {
return TDuration::Seconds(CurrentConfig.GetNodeDeletePeriod());
} else {
return NodeBrokerEpoch;
}
}

ui64 GetDrainInflight() const {
Expand Down
4 changes: 2 additions & 2 deletions ydb/core/mind/hive/tx__load_everything.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -338,9 +338,9 @@ class TTxLoadEverything : public TTransactionBase<THive> {
// it's safe to call here, because there is no any tablets in the node yet
node.BecomeDisconnected();
}
if (node.CanBeDeleted()) {
if (Self->TryToDeleteNode(&node)) {
// node is deleted from hashmap
db.Table<Schema::Node>().Key(nodeId).Delete();
Self->Nodes.erase(nodeId);
} else if (node.IsUnknown() && node.LocationAcquired) {
Self->AddRegisteredDataCentersNode(node.Location.GetDataCenterId(), node.Id);
}
Expand Down
2 changes: 1 addition & 1 deletion ydb/core/protos/config.proto
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1438,7 +1438,7 @@ message THiveConfig {
optional double MinPeriodBetweenReassign = 33 [default = 300.0]; // seconds
optional double TabletRestartWatchPeriod = 34 [default = 3600.0]; // seconds
optional double NodeRestartWatchPeriod = 35 [default = 3600.0]; // seconds
optional uint64 NodeDeletePeriod = 36 [default = 14400]; // seconds
optional uint64 NodeDeletePeriod = 36 [default = 3600]; // seconds
repeated THiveTabletLimit DefaultTabletLimit = 37;
repeated THiveTabletPreference DefaultTabletPreference = 38;
optional uint64 SystemTabletCategoryId = 39 [default = 1];
Expand Down
Loading