Skip to content

Nexus caller timeouts #9010

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
bergundy merged 3 commits into from
Jan 13, 2026
Merged

Nexus caller timeouts #9010

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4c67e93
Nexus caller timeouts
bergundy Jan 12, 2026
10c6f34
Merged API dep, affect operation timeout
bergundy Jan 13, 2026
a3b0e54
Fix lint
bergundy Jan 13, 2026
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
91 changes: 65 additions & 26 deletions api/persistence/v1/executions.pb.go
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

9 changes: 7 additions & 2 deletions components/nexusoperations/events_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,12 +11,15 @@ import (
"go.temporal.io/server/components/nexusoperations"
"go.temporal.io/server/service/history/hsm"
"go.temporal.io/server/service/history/hsm/hsmtest"
"google.golang.org/protobuf/types/known/durationpb"
"google.golang.org/protobuf/types/known/timestamppb"
)

func TestCherryPick(t *testing.T) {
setup := func(t *testing.T) (*hsm.Node, nexusoperations.Operation, int64) {
node := newOperationNode(t, &hsmtest.NodeBackend{}, mustNewScheduledEvent(time.Now(), time.Hour))
node := newOperationNode(t, &hsmtest.NodeBackend{}, mustNewScheduledEvent(time.Now(), &historypb.NexusOperationScheduledEventAttributes{
ScheduleToCloseTimeout: durationpb.New(time.Hour),
}))
op, err := hsm.MachineData[nexusoperations.Operation](node)
require.NoError(t, err)
eventID, err := hsm.EventIDFromToken(op.ScheduledEventToken)
Expand Down Expand Up @@ -159,7 +162,9 @@ func TestTerminalStatesDeletion(t *testing.T) {

for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
node := newOperationNode(t, &hsmtest.NodeBackend{}, mustNewScheduledEvent(time.Now(), time.Hour))
node := newOperationNode(t, &hsmtest.NodeBackend{}, mustNewScheduledEvent(time.Now(), &historypb.NexusOperationScheduledEventAttributes{
ScheduleToCloseTimeout: durationpb.New(time.Hour),
}))
op, err := hsm.MachineData[nexusoperations.Operation](node)
require.NoError(t, err)
eventID, err := hsm.EventIDFromToken(op.ScheduledEventToken)
Expand Down
119 changes: 108 additions & 11 deletions components/nexusoperations/executors.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,6 +74,18 @@ func RegisterExecutor(
); err != nil {
return err
}
if err := hsm.RegisterTimerExecutor(
registry,
exec.executeScheduleToStartTimeoutTask,
); err != nil {
return err
}
if err := hsm.RegisterTimerExecutor(
registry,
exec.executeStartToCloseTimeoutTask,
); err != nil {
return err
}
if err := hsm.RegisterImmediateExecutor(
registry,
exec.executeCancelationTask,
Expand Down Expand Up @@ -184,16 +196,23 @@ func (e taskExecutor) executeInvocationTask(ctx context.Context, env hsm.Environ
}

header := nexus.Header(args.header)
var opTimeout time.Duration
callTimeout := e.Config.RequestTimeout(ns.Name().String(), task.EndpointName)
if args.scheduleToCloseTimeout > 0 {
opTimeout := args.scheduleToCloseTimeout - time.Since(args.scheduledTime)
// Adjust timeout based on remaining operation timeouts.
// StartToClose takes precedence over ScheduleToClose since it is already capped by it.
if args.startToCloseTimeout > 0 {
opTimeout = args.startToCloseTimeout - time.Since(args.scheduledTime)
callTimeout = min(callTimeout, opTimeout)
if opTimeoutHeader := header.Get(nexus.HeaderOperationTimeout); opTimeoutHeader == "" {
if header == nil {
header = make(nexus.Header, 1)
}
header[nexus.HeaderOperationTimeout] = opTimeout.String()
} else if args.scheduleToCloseTimeout > 0 {
opTimeout = args.scheduleToCloseTimeout - time.Since(args.scheduledTime)
callTimeout = min(callTimeout, opTimeout)
}
// Set the operation timeout header if not already set.
if opTimeoutHeader := header.Get(nexus.HeaderOperationTimeout); opTimeout > 0 && opTimeoutHeader == "" {
if header == nil {
header = make(nexus.Header, 1)
}
header[nexus.HeaderOperationTimeout] = commonnexus.FormatDuration(opTimeout)
}

callCtx, cancel := context.WithTimeout(ctx, callTimeout)
Expand Down Expand Up @@ -311,6 +330,7 @@ type startArgs struct {
endpointID string
scheduledTime time.Time
scheduleToCloseTimeout time.Duration
startToCloseTimeout time.Duration
header map[string]string
payload *commonpb.Payload
nexusLink nexus.Link
Expand All @@ -335,15 +355,17 @@ func (e taskExecutor) loadOperationArgs(
args.service = operation.Service
args.operation = operation.Operation
args.requestID = operation.RequestId
args.scheduleToCloseTimeout = operation.ScheduleToCloseTimeout.AsDuration()
args.startToCloseTimeout = operation.StartToCloseTimeout.AsDuration()
eventToken = operation.ScheduledEventToken
event, err := node.LoadHistoryEvent(ctx, eventToken)
if err != nil {
return nil
}
args.scheduledTime = event.EventTime.AsTime()
args.scheduleToCloseTimeout = event.GetNexusOperationScheduledEventAttributes().GetScheduleToCloseTimeout().AsDuration()
args.payload = event.GetNexusOperationScheduledEventAttributes().GetInput()
args.header = event.GetNexusOperationScheduledEventAttributes().GetNexusHeader()
attrs := event.GetNexusOperationScheduledEventAttributes()
args.payload = attrs.GetInput()
args.header = attrs.GetNexusHeader()
args.nexusLink = ConvertLinkWorkflowEventToNexusLink(&commonpb.Link_WorkflowEvent{
Namespace: ns.Name().String(),
WorkflowId: ref.WorkflowKey.WorkflowID,
Expand Down Expand Up @@ -550,6 +572,74 @@ func (e taskExecutor) recordOperationTimeout(node *hsm.Node) error {
})
}

func (e taskExecutor) executeScheduleToStartTimeoutTask(env hsm.Environment, node *hsm.Node, task ScheduleToStartTimeoutTask) error {
return hsm.MachineTransition(node, func(op Operation) (hsm.TransitionOutput, error) {
eventID, err := hsm.EventIDFromToken(op.ScheduledEventToken)
if err != nil {
return hsm.TransitionOutput{}, err
}
node.AddHistoryEvent(enumspb.EVENT_TYPE_NEXUS_OPERATION_TIMED_OUT, func(e *historypb.HistoryEvent) {
// nolint:revive // We must mutate here even if the linter doesn't like it.
e.Attributes = &historypb.HistoryEvent_NexusOperationTimedOutEventAttributes{
NexusOperationTimedOutEventAttributes: &historypb.NexusOperationTimedOutEventAttributes{
Failure: nexusOperationFailure(
op,
eventID,
&failurepb.Failure{
Message: "operation timed out before starting",
FailureInfo: &failurepb.Failure_TimeoutFailureInfo{
TimeoutFailureInfo: &failurepb.TimeoutFailureInfo{
TimeoutType: enumspb.TIMEOUT_TYPE_SCHEDULE_TO_START,
},
},
},
),
ScheduledEventId: eventID,
RequestId: op.RequestId,
},
}
})

return TransitionTimedOut.Apply(op, EventTimedOut{
Node: node,
})
})
}

func (e taskExecutor) executeStartToCloseTimeoutTask(env hsm.Environment, node *hsm.Node, task StartToCloseTimeoutTask) error {
return hsm.MachineTransition(node, func(op Operation) (hsm.TransitionOutput, error) {
eventID, err := hsm.EventIDFromToken(op.ScheduledEventToken)
if err != nil {
return hsm.TransitionOutput{}, err
}
node.AddHistoryEvent(enumspb.EVENT_TYPE_NEXUS_OPERATION_TIMED_OUT, func(e *historypb.HistoryEvent) {
// nolint:revive // We must mutate here even if the linter doesn't like it.
e.Attributes = &historypb.HistoryEvent_NexusOperationTimedOutEventAttributes{
NexusOperationTimedOutEventAttributes: &historypb.NexusOperationTimedOutEventAttributes{
Failure: nexusOperationFailure(
op,
eventID,
&failurepb.Failure{
Message: "operation timed out after starting",
FailureInfo: &failurepb.Failure_TimeoutFailureInfo{
TimeoutFailureInfo: &failurepb.TimeoutFailureInfo{
TimeoutType: enumspb.TIMEOUT_TYPE_START_TO_CLOSE,
},
},
},
),
ScheduledEventId: eventID,
RequestId: op.RequestId,
},
}
})

return TransitionTimedOut.Apply(op, EventTimedOut{
Node: node,
})
})
}

func (e taskExecutor) executeCancelationTask(ctx context.Context, env hsm.Environment, ref hsm.Ref, task CancelationTask) error {
ns, err := e.NamespaceRegistry.GetNamespaceByID(namespace.ID(ref.WorkflowKey.NamespaceID))
if err != nil {
Expand All @@ -573,7 +663,12 @@ func (e taskExecutor) executeCancelationTask(ctx context.Context, env hsm.Enviro
}

callTimeout := e.Config.RequestTimeout(ns.Name().String(), task.EndpointName)
if args.scheduleToCloseTimeout > 0 {
// Adjust timeout based on remaining operation timeouts.
// StartToClose takes precedence over ScheduleToClose since it is already capped by it.
if args.startToCloseTimeout > 0 {
opTimeout := args.startToCloseTimeout - time.Since(args.scheduledTime)
callTimeout = min(callTimeout, opTimeout)
} else if args.scheduleToCloseTimeout > 0 {
opTimeout := args.scheduleToCloseTimeout - time.Since(args.scheduledTime)
callTimeout = min(callTimeout, opTimeout)
}
Expand Down Expand Up @@ -652,6 +747,7 @@ type cancelArgs struct {
service, operation, token, endpointID, endpointName, requestID string
scheduledTime time.Time
scheduleToCloseTimeout time.Duration
startToCloseTimeout time.Duration
scheduledEventID int64
headers map[string]string
}
Expand All @@ -677,6 +773,7 @@ func (e taskExecutor) loadArgsForCancelation(ctx context.Context, env hsm.Enviro
args.requestID = op.RequestId
args.scheduledTime = op.ScheduledTime.AsTime()
args.scheduleToCloseTimeout = op.ScheduleToCloseTimeout.AsDuration()
args.startToCloseTimeout = op.StartToCloseTimeout.AsDuration()
args.scheduledEventID, err = hsm.EventIDFromToken(op.ScheduledEventToken)
if err != nil {
return err
Expand Down
Loading
Loading