mirror of
https://github.com/hanzoai/tasks.git
synced 2026-10-04 02:04:03 +00:00
## What changed? - Add "ephemeral data" propagated alongside user data. - Include information about which priority levels have significant backlog in ephemeral data (currently only propagated down the tree, not up) - When sticky partitions see that normal partitions have significant backlog, set up poll forwarders to forward to those partitions. - The poll forwarders use min priority to ensure they only get higher-priority tasks than available local tasks. ## Why? Without this functionality, sticky queues interfere with priority dispatch by keeping most of the pollers working on sticky tasks and potentially starving high-priority tasks that appear on the normal queue. With the default values of the new settings, sticky pollers should notice high-priority normal tasks within 10 seconds. ## How did you test it? - [ ] built - [ ] run locally and tested manually - [ ] covered by existing tests - [x] added new unit test(s) - [x] added new functional test(s) ## Potential risks - Clients have to accept that polls on sticky may return normal tasks. I believe all SDKs are fine with this. - Increased overhead of user data + ephemeral data.
281 lines
9.6 KiB
Go
281 lines
9.6 KiB
Go
package matching
|
|
|
|
import (
|
|
"context"
|
|
"time"
|
|
|
|
enumspb "go.temporal.io/api/enums/v1"
|
|
"go.temporal.io/api/serviceerror"
|
|
taskqueuepb "go.temporal.io/api/taskqueue/v1"
|
|
"go.temporal.io/api/workflowservice/v1"
|
|
"go.temporal.io/server/api/matchingservice/v1"
|
|
taskqueuespb "go.temporal.io/server/api/taskqueue/v1"
|
|
"go.temporal.io/server/common"
|
|
"go.temporal.io/server/common/tqid"
|
|
"google.golang.org/protobuf/types/known/durationpb"
|
|
)
|
|
|
|
type (
|
|
// priForwarder is the type that contains state pertaining to
|
|
// the api call forwarder component
|
|
priForwarder struct {
|
|
cfg *forwarderConfig
|
|
queue *PhysicalTaskQueueKey
|
|
partition *tqid.NormalPartition
|
|
client matchingservice.MatchingServiceClient
|
|
}
|
|
)
|
|
|
|
// TODO(pri): old matcher cleanup, move to here
|
|
// var errInvalidTaskQueueType = errors.New("unrecognized task queue type")
|
|
|
|
// newPriForwarder returns an instance of priForwarder object which
|
|
// can be used to forward api request calls from a task queue
|
|
// child partition to a task queue parent partition. The returned
|
|
// forwarder is tied to a single task queue partition. All exposed
|
|
// methods can return the following errors:
|
|
// Returns following errors:
|
|
// - taskqueue.ErrNoParent, taskqueue.ErrInvalidDegree: If this task queue doesn't have a parent to forward to
|
|
func newPriForwarder(
|
|
cfg *forwarderConfig,
|
|
queue *PhysicalTaskQueueKey,
|
|
client matchingservice.MatchingServiceClient,
|
|
) (*priForwarder, error) {
|
|
partition, ok := queue.Partition().(*tqid.NormalPartition)
|
|
if !ok {
|
|
return nil, serviceerror.NewInvalidArgument("physical queue of normal partition expected")
|
|
}
|
|
return &priForwarder{
|
|
cfg: cfg,
|
|
client: client,
|
|
partition: partition,
|
|
queue: queue,
|
|
}, nil
|
|
}
|
|
|
|
// ForwardTask forwards an activity or workflow task to the parent task queue partition if it exists
|
|
func (f *priForwarder) ForwardTask(ctx context.Context, task *internalTask) error {
|
|
degree := f.cfg.ForwarderMaxChildrenPerNode()
|
|
target, err := f.partition.ParentPartition(degree)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
var expirationDuration *durationpb.Duration
|
|
var expirationTime time.Time
|
|
if task.event.Data.ExpiryTime != nil {
|
|
expirationTime = task.event.Data.ExpiryTime.AsTime()
|
|
remaining := time.Until(expirationTime)
|
|
if remaining <= 0 {
|
|
return nil
|
|
}
|
|
expirationDuration = durationpb.New(remaining)
|
|
}
|
|
|
|
// nolint:exhaustive // there's a default clause
|
|
switch f.partition.TaskType() {
|
|
case enumspb.TASK_QUEUE_TYPE_WORKFLOW:
|
|
_, err = f.client.AddWorkflowTask(
|
|
ctx, &matchingservice.AddWorkflowTaskRequest{
|
|
NamespaceId: task.event.Data.GetNamespaceId(),
|
|
Execution: task.workflowExecution(),
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: f.partition.Kind(),
|
|
},
|
|
ScheduledEventId: task.event.Data.GetScheduledEventId(),
|
|
Clock: task.event.Data.GetClock(),
|
|
ScheduleToStartTimeout: expirationDuration,
|
|
ForwardInfo: f.getForwardInfo(task),
|
|
VersionDirective: task.event.Data.GetVersionDirective(),
|
|
Stamp: task.event.Data.GetStamp(),
|
|
Priority: task.event.Data.GetPriority(),
|
|
},
|
|
)
|
|
case enumspb.TASK_QUEUE_TYPE_ACTIVITY:
|
|
_, err = f.client.AddActivityTask(
|
|
ctx, &matchingservice.AddActivityTaskRequest{
|
|
NamespaceId: task.event.Data.GetNamespaceId(),
|
|
Execution: task.workflowExecution(),
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: f.partition.Kind(),
|
|
},
|
|
ScheduledEventId: task.event.Data.GetScheduledEventId(),
|
|
Clock: task.event.Data.GetClock(),
|
|
ScheduleToStartTimeout: expirationDuration,
|
|
ForwardInfo: f.getForwardInfo(task),
|
|
VersionDirective: task.event.Data.GetVersionDirective(),
|
|
Stamp: task.event.Data.GetStamp(),
|
|
Priority: task.event.Data.GetPriority(),
|
|
ComponentRef: task.event.Data.GetComponentRef(),
|
|
},
|
|
)
|
|
default:
|
|
return errInvalidTaskQueueType
|
|
}
|
|
|
|
return err
|
|
}
|
|
|
|
func (f *priForwarder) getForwardInfo(task *internalTask) *taskqueuespb.TaskForwardInfo {
|
|
if task.isForwarded() {
|
|
// task is already forwarded from a child partition, only overwrite SourcePartition
|
|
clone := common.CloneProto(task.forwardInfo)
|
|
clone.SourcePartition = f.partition.RpcName()
|
|
return clone
|
|
}
|
|
// task is forwarded for the first time
|
|
return &taskqueuespb.TaskForwardInfo{
|
|
TaskSource: task.source,
|
|
SourcePartition: f.partition.RpcName(),
|
|
DispatchBuildId: f.queue.Version().BuildId(),
|
|
DispatchVersionSet: f.queue.Version().VersionSet(),
|
|
RedirectInfo: task.redirectInfo,
|
|
}
|
|
}
|
|
|
|
// ForwardQueryTask forwards a query task to parent task queue partition, if it exists
|
|
func (f *priForwarder) ForwardQueryTask(
|
|
ctx context.Context,
|
|
task *internalTask,
|
|
) (*matchingservice.QueryWorkflowResponse, error) {
|
|
degree := f.cfg.ForwarderMaxChildrenPerNode()
|
|
target, err := f.partition.ParentPartition(degree)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
resp, err := f.client.QueryWorkflow(ctx, &matchingservice.QueryWorkflowRequest{
|
|
NamespaceId: task.query.request.GetNamespaceId(),
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: f.partition.Kind(),
|
|
},
|
|
QueryRequest: task.query.request.QueryRequest,
|
|
VersionDirective: task.query.request.VersionDirective,
|
|
ForwardInfo: f.getForwardInfo(task),
|
|
})
|
|
|
|
return resp, err
|
|
}
|
|
|
|
// ForwardNexusTask forwards a nexus task to parent task queue partition, if it exists.
|
|
func (f *priForwarder) ForwardNexusTask(ctx context.Context, task *internalTask) (*matchingservice.DispatchNexusTaskResponse, error) {
|
|
degree := f.cfg.ForwarderMaxChildrenPerNode()
|
|
target, err := f.partition.ParentPartition(degree)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
resp, err := f.client.DispatchNexusTask(ctx, &matchingservice.DispatchNexusTaskRequest{
|
|
NamespaceId: task.nexus.request.GetNamespaceId(),
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: f.partition.Kind(),
|
|
},
|
|
Request: task.nexus.request.Request,
|
|
ForwardInfo: f.getForwardInfo(task),
|
|
})
|
|
|
|
return resp, err
|
|
}
|
|
|
|
// ForwardPoll forwards a poll request to parent task queue partition if it exist
|
|
// TODO(pri): remove this and update tests to call ForwardPollWithTarget directly
|
|
func (f *priForwarder) ForwardPoll(ctx context.Context, pollMetadata *pollMetadata) (*internalTask, error) {
|
|
degree := f.cfg.ForwarderMaxChildrenPerNode()
|
|
target, err := f.partition.ParentPartition(degree)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
return ForwardPollWithTarget(ctx, pollMetadata, f.client, f.partition, target)
|
|
}
|
|
|
|
// ForwardPollWithTarget forwards a poll request to another partition
|
|
func ForwardPollWithTarget(
|
|
ctx context.Context,
|
|
pollMetadata *pollMetadata,
|
|
client matchingservice.MatchingServiceClient,
|
|
source tqid.Partition,
|
|
target *tqid.NormalPartition,
|
|
) (*internalTask, error) {
|
|
pollerID, _ := ctx.Value(pollerIDKey).(string) // nolint:revive
|
|
identity, _ := ctx.Value(identityKey).(string) // nolint:revive
|
|
|
|
// nolint:exhaustive // there's a default clause
|
|
switch target.TaskType() {
|
|
case enumspb.TASK_QUEUE_TYPE_WORKFLOW:
|
|
resp, err := client.PollWorkflowTaskQueue(ctx, &matchingservice.PollWorkflowTaskQueueRequest{
|
|
NamespaceId: target.TaskQueue().NamespaceId(),
|
|
PollerId: pollerID,
|
|
PollRequest: &workflowservice.PollWorkflowTaskQueueRequest{
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: enumspb.TASK_QUEUE_KIND_NORMAL,
|
|
},
|
|
Identity: identity,
|
|
WorkerVersionCapabilities: pollMetadata.workerVersionCapabilities,
|
|
DeploymentOptions: pollMetadata.deploymentOptions,
|
|
},
|
|
ForwardedSource: source.RpcName(),
|
|
Conditions: pollMetadata.conditions,
|
|
})
|
|
if err != nil {
|
|
return nil, err
|
|
} else if resp.TaskToken == nil {
|
|
return nil, errNoTasks
|
|
}
|
|
return newInternalStartedTask(&startedTaskInfo{workflowTaskInfo: resp}), nil
|
|
case enumspb.TASK_QUEUE_TYPE_ACTIVITY:
|
|
resp, err := client.PollActivityTaskQueue(ctx, &matchingservice.PollActivityTaskQueueRequest{
|
|
NamespaceId: target.TaskQueue().NamespaceId(),
|
|
PollerId: pollerID,
|
|
PollRequest: &workflowservice.PollActivityTaskQueueRequest{
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: enumspb.TASK_QUEUE_KIND_NORMAL,
|
|
},
|
|
Identity: identity,
|
|
TaskQueueMetadata: pollMetadata.taskQueueMetadata,
|
|
WorkerVersionCapabilities: pollMetadata.workerVersionCapabilities,
|
|
DeploymentOptions: pollMetadata.deploymentOptions,
|
|
},
|
|
ForwardedSource: source.RpcName(),
|
|
Conditions: pollMetadata.conditions,
|
|
})
|
|
if err != nil {
|
|
return nil, err
|
|
} else if resp.TaskToken == nil {
|
|
return nil, errNoTasks
|
|
}
|
|
return newInternalStartedTask(&startedTaskInfo{activityTaskInfo: resp}), nil
|
|
case enumspb.TASK_QUEUE_TYPE_NEXUS:
|
|
resp, err := client.PollNexusTaskQueue(ctx, &matchingservice.PollNexusTaskQueueRequest{
|
|
NamespaceId: target.TaskQueue().NamespaceId(),
|
|
PollerId: pollerID,
|
|
Request: &workflowservice.PollNexusTaskQueueRequest{
|
|
TaskQueue: &taskqueuepb.TaskQueue{
|
|
Name: target.RpcName(),
|
|
Kind: enumspb.TASK_QUEUE_KIND_NORMAL,
|
|
},
|
|
Identity: identity,
|
|
WorkerVersionCapabilities: pollMetadata.workerVersionCapabilities,
|
|
DeploymentOptions: pollMetadata.deploymentOptions,
|
|
// Namespace is ignored here.
|
|
},
|
|
ForwardedSource: source.RpcName(),
|
|
Conditions: pollMetadata.conditions,
|
|
})
|
|
if err != nil {
|
|
return nil, err
|
|
} else if resp.Response == nil {
|
|
return nil, errNoTasks
|
|
}
|
|
return newInternalStartedTask(&startedTaskInfo{nexusTaskInfo: resp}), nil
|
|
default:
|
|
return nil, errInvalidTaskQueueType
|
|
}
|
|
}
|