Skip to content

Improve KqpHost introspection, add KqpExecuter and CA virtual pages #30205

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
Hor911 merged 1 commit into from
Dec 8, 2025
+173 −22
Merged

Improve KqpHost introspection, add KqpExecuter and CA virtual pages #30205

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
1 change: 1 addition & 0 deletions ydb/core/kqp/executer_actor/kqp_data_executer.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1142,6 +1142,7 @@ class TKqpDataExecuter : public TKqpExecuterBase<TKqpDataExecuter, EExecType::Da
hFunc(TEvKqpBuffer::TEvError, Handle);
hFunc(NFq::TEvCheckpointCoordinator::TEvZeroCheckpointDone, Handle);
hFunc(NFq::TEvCheckpointCoordinator::TEvRaiseTransientIssues, Handle);
hFunc(NActors::NMon::TEvHttpInfo, HandleHttpInfo);
IgnoreFunc(TEvInterconnect::TEvNodeConnected);
default:
UnexpectedEvent("ExecuteState", ev->GetTypeRewrite());
Expand Down
45 changes: 45 additions & 0 deletions ydb/core/kqp/executer_actor/kqp_executer_impl.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -581,6 +581,51 @@ class TKqpExecuterBase : public TActor<TDerived> {
static_cast<TDerived*>(this)->CheckExecutionComplete();
}

void HandleHttpInfo(NMon::TEvHttpInfo::TPtr& ev) {
TStringStream str;
HTML(str) {
PRE() {
str << "KQP Executer, SelfId=" << SelfId() << Endl;

TABLE_SORTABLE_CLASS("table table-condensed") {
TABLEHEAD() {
TABLER() {
TABLEH() {str << "TxId";}
TABLEH() {str << "StageId";}
TABLEH() {str << "TaskId";}
TABLEH() {str << "NodeId";}
TABLEH() {str << "ActorId";}
TABLEH() {str << "Completed";}
}
}
TABLEBODY() {
for (const auto& task : TasksGraph.GetTasks()) {
TABLER() {
TABLED() {str << task.StageId.TxId;}
TABLED() {str << task.StageId.StageId;}
TABLED() {str << task.Id;}
TABLED() {str << task.Meta.NodeId;}
TABLED() {
if (task.ComputeActorId) {
HREF(TStringBuilder() << "/node/" << task.ComputeActorId.NodeId() << "/actors/kqp_node?ca=" << task.ComputeActorId) {
str << task.ComputeActorId;
}
} else {
str << "N/A";
}
str << Endl;
}
TABLED() {str << task.Meta.Completed;}
}
}
}
}
}
}

this->Send(ev->Sender, new NMon::TEvHttpInfoRes(str.Str()));
}

STATEFN(ReadyState) {
switch (ev->GetTypeRewrite()) {
hFunc(TEvKqpExecuter::TEvTxRequest, HandleReady);
Expand Down
1 change: 1 addition & 0 deletions ydb/core/kqp/executer_actor/kqp_scan_executer.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,6 +105,7 @@ class TKqpScanExecuter : public TKqpExecuterBase<TKqpScanExecuter, EExecType::Sc
hFunc(TEvKqpNode::TEvStartKqpTasksResponse, HandleStartKqpTasksResponse);
IgnoreFunc(TEvKqpNode::TEvCancelKqpTasksResponse);
hFunc(TEvInterconnect::TEvNodeDisconnected, HandleDisconnected);
hFunc(NActors::NMon::TEvHttpInfo, HandleHttpInfo);
IgnoreFunc(TEvInterconnect::TEvNodeConnected);
default:
UnexpectedEvent("ExecuteState", ev->GetTypeRewrite());
Expand Down
22 changes: 18 additions & 4 deletions ydb/core/kqp/node_service/kqp_node_service.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -510,14 +510,28 @@ class TKqpNodeService : public TActorBootstrapped<TKqpNodeService> {
}

void HandleWork(NMon::TEvHttpInfo::TPtr& ev) {

const TCgiParameters &cgi = ev->Get()->Request.GetParams();
TActorId id;

auto caId = cgi.Get("ca");
if (caId && State_->ValidateComputeActorId(caId, id)) {
TActivationContext::Send(ev->Forward(id));
return;
}

auto exId = cgi.Get("ex");
if (exId && State_->ValidateKqpExecuterId(exId, SelfId().NodeId(), id)) {
TActivationContext::Send(ev->Forward(id));
return;
}

TStringStream str;
HTML(str) {
PRE() {
str << "Current config:" << Endl;
str << "TKqpNodeService, SelfId=" << SelfId() << Endl;
str << Endl << "Current config:" << Endl;
str << Config.DebugString() << Endl;
str << Endl;

str << Endl << "Transactions:" << Endl;
State_->DumpInfo(str);
}
}
Expand Down
124 changes: 106 additions & 18 deletions ydb/core/kqp/node_service/kqp_node_state.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,31 +113,119 @@ std::vector<TNodeRequest::TTaskInfo> TNodeState::GetTasksByTxId(ui64 txId) const
}

void TNodeState::DumpInfo(TStringStream& str) const {
for (const auto& bucket : Buckets) {
TReadGuard guard(bucket.Mutex);
TMap<ui64, TVector<std::pair<const TActorId, const TNodeRequest*>>> byTx;
HTML(str) {
str << Endl << "Transactions:" << Endl;
TABLE_SORTABLE_CLASS("table table-condensed") {
TABLEHEAD() {
TABLER() {
TABLEH() {str << "TxId";}
TABLEH() {str << "Executer";}
TABLEH() {str << "StartTime";}
TABLEH() {str << "Deadline";}
}
}
TABLEBODY() {
for (const auto& bucket : Buckets) {
TReadGuard guard(bucket.Mutex);
TMap<ui64, TVector<std::pair<const TActorId, const TNodeRequest*>>> byTx;

for (const auto& [txId, request] : bucket.Requests) {
byTx[txId].emplace_back(request.ExecuterId, &request);
for (const auto& [txId, request] : bucket.Requests) {
byTx[txId].emplace_back(request.ExecuterId, &request);
}

for (const auto& [txId, requests] : byTx) {
for (auto& [requester, request] : requests) {
TABLER() {
TABLED() {str << txId;}
TABLED() {
HREF(TStringBuilder() << "/node/" << requester.NodeId() << "/actors/kqp_node?ex=" << requester) {
str << requester;
}
}
TABLED() {str << request->StartTime;}
TABLED() {str << request->Deadline;}
}
}
}
}
}
}
for (const auto& [txId, requests] : byTx) {
str << " Requests:" << Endl;
for (auto& [requester, request] : requests) {
str << " Requester: " << requester << Endl;
str << " StartTime: " << request->StartTime << Endl;
str << " Deadline: " << request->Deadline << Endl;
str << " In-fly tasks:" << Endl;
for (auto& [taskId, actorId] : request->Tasks) {
str << " Task: " << taskId << Endl;
if (actorId) {
str << " Compute actor: " << *actorId << Endl;
} else {
str << " Compute actor: (task not started yet)" << Endl;

str << Endl << "Tasks:" << Endl;
TABLE_SORTABLE_CLASS("table table-condensed") {
TABLEHEAD() {
TABLER() {
TABLEH() {str << "TxId";}
TABLEH() {str << "Executer";}
TABLEH() {str << "TaskId";}
TABLEH() {str << "ComputeActorId";}
}
}
TABLEBODY() {
for (const auto& bucket : Buckets) {
TReadGuard guard(bucket.Mutex);
TMap<ui64, TVector<std::pair<const TActorId, const TNodeRequest*>>> byTx;

for (const auto& [txId, request] : bucket.Requests) {
byTx[txId].emplace_back(request.ExecuterId, &request);
}

for (const auto& [txId, requests] : byTx) {
for (auto& [requester, request] : requests) {
for (auto& [taskId, actorId] : request->Tasks) {
TABLER() {
TABLED() {str << txId;}
TABLED() {
HREF(TStringBuilder() << "/node/" << requester.NodeId() << "/actors/kqp_node?ex=" << requester) {
str << requester;
}
}
TABLED() {str << taskId;}
TABLED() {
if (actorId) {
HREF(TStringBuilder() << "/node/" << requester.NodeId() << "/actors/kqp_node?ca=" << *actorId) {
str << *actorId;
}
} else {
str << "N/A";
}
}
}
}
}
}
}
}
}
}
}

bool TNodeState::ValidateComputeActorId(const TString& computeActorId, TActorId& id) const {
for (const auto& bucket : Buckets) {
TReadGuard guard(bucket.Mutex);
for (const auto& [_, request] : bucket.Requests) {
for (auto& [_, actorId] : request.Tasks) {
if (actorId && ToString(*actorId) == computeActorId) {
id = *actorId;
return true;
}
}
}
}
return false;
}

bool TNodeState::ValidateKqpExecuterId(const TString& kqpExecuterId, ui32 nodeId, TActorId& id) const {
for (const auto& bucket : Buckets) {
TReadGuard guard(bucket.Mutex);
for (const auto& [_, request] : bucket.Requests) {
if (ToString(request.ExecuterId) == kqpExecuterId && request.ExecuterId.NodeId() == nodeId) {
id = request.ExecuterId;
return true;
}
}
}
return false;
}

} // namespace NKikimr::NKqp::NKqpNode
2 changes: 2 additions & 0 deletions ydb/core/kqp/node_service/kqp_node_state.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,6 +61,8 @@ class TNodeState {
std::vector<TNodeRequest::TTaskInfo> GetTasksByTxId(ui64 txId) const;

void DumpInfo(TStringStream& str) const;
bool ValidateComputeActorId(const TString& computeActorId, TActorId& id) const;
bool ValidateKqpExecuterId(const TString& kqpExecuterId, ui32 nodeId, TActorId& id) const;

private:
inline auto& GetBucketByTxId(ui64 txId) {
Expand Down
Loading