Documentation
¶
Overview ¶
Package scheduler ensures all tasks in a pipeline and child pipelines are loaded and executed in the order they need to be, parallelizing where possible.
Index ¶
- Constants
- Variables
- type ExecutionGraph
- func (g *ExecutionGraph) AddStage(stage *Stage) error
- func (g *ExecutionGraph) BFSNodesFlattened(nodeName string) []*Stage
- func (g *ExecutionGraph) Children(node string) map[string]*Stage
- func (g *ExecutionGraph) Duration() time.Duration
- func (g *ExecutionGraph) From(name string) []string
- func (g *ExecutionGraph) HasCycle() error
- func (g *ExecutionGraph) LastError() error
- func (g *ExecutionGraph) Name() string
- func (g *ExecutionGraph) Node(name string) (*Stage, error)
- func (g *ExecutionGraph) Nodes() map[string]*Stage
- func (g *ExecutionGraph) To(name string) []string
- type Scheduler
- type Stage
- type StageOpts
Examples ¶
Constants ¶
const ( StatusWaiting int32 = iota StatusRunning StatusSkipped StatusDone StatusError StatusCanceled )
Stage statuses
const (
RootNodeName = "root"
)
Variables ¶
var ErrCycleDetected = errors.New("cycle detected")
ErrCycleDetected occurs when added edge causes cycle to appear
Functions ¶
This section is empty.
Types ¶
type ExecutionGraph ¶
type ExecutionGraph struct {
Generator map[string]any
Env map[string][]string
// contains filtered or unexported fields
}
ExecutionGraph is a DAG whose nodes are Stages and edges are their dependencies
func NewExecutionGraph ¶
func NewExecutionGraph(name string, stages ...*Stage) (*ExecutionGraph, error)
NewExecutionGraph creates new ExecutionGraph instance. It accepts zero or more stages and adds them to resulted graph
func (*ExecutionGraph) AddStage ¶
func (g *ExecutionGraph) AddStage(stage *Stage) error
AddStage adds Stage to ExecutionGraph. If newly added stage causes a cycle to appear in the graph it return an error
func (*ExecutionGraph) BFSNodesFlattened ¶
func (g *ExecutionGraph) BFSNodesFlattened(nodeName string) []*Stage
BFSNodesFlattened returns a Breadth-First-Search flattened list of top level tasks/pipelines This is useful in summaries as we want the things that run in parallel on the same level to show in that order before the level below and so on.
When generating CI definitions - we don't need to generate the same jobs/steps over and over again they will be referenced with a needs/depends_on/etc... keyword
func (*ExecutionGraph) Duration ¶
func (g *ExecutionGraph) Duration() time.Duration
Duration returns execution duration
func (*ExecutionGraph) From ¶
func (g *ExecutionGraph) From(name string) []string
From returns stages that depend on the given stage Children of the node
func (*ExecutionGraph) HasCycle ¶
func (g *ExecutionGraph) HasCycle() error
func (*ExecutionGraph) LastError ¶
func (g *ExecutionGraph) LastError() error
LastError returns latest error appeared during stages execution
func (*ExecutionGraph) Name ¶
func (g *ExecutionGraph) Name() string
LastError returns latest error appeared during stages execution
func (*ExecutionGraph) Node ¶
func (g *ExecutionGraph) Node(name string) (*Stage, error)
Node returns stage by its name
func (*ExecutionGraph) Nodes ¶
func (g *ExecutionGraph) Nodes() map[string]*Stage
Nodes returns ExecutionGraph stages - an n-ary tree itself Stage (Node) may appear multiple times in a scheduling scenario, this is desired behaviour to loop over the nodes as many times as they appear in a DAG manner.
func (*ExecutionGraph) To ¶
func (g *ExecutionGraph) To(name string) []string
To returns stages on whiсh given stage depends on returns the parents of the node
type Scheduler ¶
type Scheduler struct {
// contains filtered or unexported fields
}
Scheduler executes ExecutionGraph
func NewScheduler ¶
NewScheduler create new Scheduler instance
func (*Scheduler) Cancel ¶
func (s *Scheduler) Cancel()
Cancel cancels executing tasks atomically loads the cancelled code
func (*Scheduler) Schedule ¶
func (s *Scheduler) Schedule(g *ExecutionGraph) error
Schedule starts execution of the given ExecutionGraph
Example ¶
package main
import (
"fmt"
"github.com/Ensono/taskctl/pkg/scheduler"
"github.com/Ensono/taskctl/pkg/runner"
"github.com/Ensono/taskctl/pkg/task"
)
func main() {
format := task.FromCommands("t1", "go fmt ./...")
build := task.FromCommands("t2", "go build ./..")
r, _ := runner.NewTaskRunner()
s := scheduler.NewScheduler(r)
graph, err := scheduler.NewExecutionGraph("t1",
scheduler.NewStage(func(s *scheduler.Stage) {
s.Name = "format"
s.Task = format
}),
scheduler.NewStage(func(s *scheduler.Stage) {
s.Name = "build"
s.Task = build
s.DependsOn = []string{"format"}
}),
)
if err != nil {
return
}
err = s.Schedule(graph)
if err != nil {
fmt.Println(err)
}
}
type Stage ¶
type Stage struct {
Name string
Condition string
Task *task.Task
Pipeline *ExecutionGraph
DependsOn []string
Dir string
AllowFailure bool
Status *atomic.Int32
Env variables.Container
Variables variables.Container
Start time.Time
End time.Time
Generator map[string]any
}
Stage is a structure that describes execution stage Stage is a synonym for a Node in a the unary tree of the execution graph/tree
func (*Stage) ReadStatus ¶
ReadStatus is a helper to read stage's status atomically
func (*Stage) UpdateStatus ¶
UpdateStatus updates stage's status atomically
Source Files