359 lines
11 KiB
Go
359 lines
11 KiB
Go
package controllers
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"net/http"
|
|
"oc-scheduler/infrastructure"
|
|
"reflect"
|
|
"strconv"
|
|
"strings"
|
|
"sync"
|
|
"time"
|
|
|
|
oclib "cloud.o-forge.io/core/oc-lib"
|
|
"cloud.o-forge.io/core/oc-lib/dbs"
|
|
"cloud.o-forge.io/core/oc-lib/tools"
|
|
beego "github.com/beego/beego/v2/server/web"
|
|
"github.com/google/uuid"
|
|
gorillaws "github.com/gorilla/websocket"
|
|
)
|
|
|
|
var orderCollection = oclib.LibDataEnum(oclib.ORDER)
|
|
var logger = oclib.GetLogger()
|
|
|
|
// Operations about workflow
|
|
type WorkflowSchedulerController struct {
|
|
beego.Controller
|
|
}
|
|
|
|
var wsUpgrader = gorillaws.Upgrader{
|
|
CheckOrigin: func(r *http.Request) bool { return true },
|
|
}
|
|
|
|
var schedulerMu sync.RWMutex
|
|
var scheduler = map[string]*infrastructure.WorkflowSchedule{}
|
|
|
|
func realPushCheckfunc(ctx context.Context, conn *gorillaws.Conn, req *tools.APIRequest, user string, ws infrastructure.WorkflowSchedule,
|
|
executionsID string, wfID string, scheduled bool, asap bool, preemption bool, reschedule bool) (bool, error) {
|
|
// If we already have draft bookings for this session and we're about to
|
|
// re-check (timer refresh or planner update), remove the old drafts first
|
|
// so the planner doesn't treat our own previous reservations as conflicts.
|
|
if reschedule && scheduled {
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
scheduled = false
|
|
}
|
|
workflowScheduler := ws
|
|
|
|
schedulerMu.Lock()
|
|
if scheduler[user] != nil {
|
|
workflowScheduler = *scheduler[user]
|
|
}
|
|
schedulerMu.Unlock()
|
|
result, checkErr := workflowScheduler.Check(wfID, asap, preemption, req)
|
|
fmt.Println("CHECK", checkErr)
|
|
if checkErr != nil {
|
|
return scheduled, checkErr
|
|
}
|
|
if result.Available && reschedule {
|
|
workflowScheduler.Start = result.Start
|
|
if result.End != nil {
|
|
workflowScheduler.End = result.End
|
|
}
|
|
_, _, execs, purchases, bookings, err := workflowScheduler.GetBuyAndBook(wfID, req)
|
|
if err != nil {
|
|
fmt.Println("GetBuyAndBook", err)
|
|
return scheduled, err
|
|
}
|
|
infrastructure.UpsertSessionDrafts(executionsID, execs, purchases, bookings, req)
|
|
scheduled = true
|
|
delay := workflowScheduler.Start.UTC().Add(-(1 * time.Minute)).Sub(time.Now().UTC())
|
|
go func() {
|
|
select {
|
|
case <-ctx.Done():
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
// Session closed before timer fired — nothing to do, CleanupSession
|
|
// has already run (or will run) in the defer of CheckStreamHandler.
|
|
return
|
|
case <-time.After(delay):
|
|
realPushCheckfunc(ctx, conn, req, user, ws, executionsID, wfID, scheduled, asap, preemption, true)
|
|
}
|
|
}()
|
|
}
|
|
result.SchedulingID = executionsID
|
|
fmt.Println(result)
|
|
return scheduled, conn.WriteJSON(result)
|
|
}
|
|
|
|
// CheckStreamHandler is the WebSocket handler for slot availability checking.
|
|
// Query params: as_possible=true, preemption=true
|
|
func CheckStreamHandler(w http.ResponseWriter, r *http.Request) {
|
|
fmt.Println("qksbdkqbsdkh")
|
|
var err error
|
|
wfID := strings.TrimSuffix(
|
|
strings.TrimPrefix(r.URL.Path, "/oc/check/"),
|
|
"",
|
|
)
|
|
|
|
q := r.URL.Query()
|
|
asap := q.Get("as_possible") == "true"
|
|
preemption := q.Get("preemption") == "true"
|
|
|
|
user, peerID, groups := oclib.ExtractTokenInfoWs(*r)
|
|
req := &tools.APIRequest{
|
|
Username: user,
|
|
PeerID: peerID,
|
|
Groups: groups,
|
|
Caller: nil,
|
|
Admin: true,
|
|
}
|
|
|
|
watchedPeers, err := infrastructure.GetWorkflowPeerIDs(wfID, req)
|
|
fmt.Println("Watched peers for workflow", wfID, ":", watchedPeers, err)
|
|
if err != nil {
|
|
http.Error(w, `{"code":404,"error":"`+err.Error()+`"}`, http.StatusNotFound)
|
|
return
|
|
}
|
|
|
|
conn, err := wsUpgrader.Upgrade(w, r, nil)
|
|
fmt.Println("Upgrade :", err)
|
|
if err != nil {
|
|
return
|
|
}
|
|
|
|
var ws infrastructure.WorkflowSchedule
|
|
if err := conn.ReadJSON(&ws); err != nil {
|
|
fmt.Println("ReadJSON :", err)
|
|
conn.Close()
|
|
return
|
|
}
|
|
// Allow the initial JSON to override the query-param mode.
|
|
if ws.Asap != nil {
|
|
asap = *ws.Asap
|
|
}
|
|
if ws.Preemption != nil {
|
|
preemption = *ws.Preemption
|
|
}
|
|
|
|
plannerCh, plannerUnsub := infrastructure.SubscribePlannerUpdates(watchedPeers)
|
|
wfCh, wfUnsub := infrastructure.SubscribeWorkflowUpdates(wfID)
|
|
|
|
executionsID := uuid.New().String()
|
|
ownedPeers := infrastructure.RequestPlannerRefresh(watchedPeers, executionsID)
|
|
|
|
self, err := oclib.GetMySelf()
|
|
if err != nil || self == nil {
|
|
logger.Err(err).Msg("could not resolve self peer")
|
|
conn.Close()
|
|
return
|
|
}
|
|
selfPeerID := self.PeerID
|
|
|
|
scheduled := false
|
|
confirmed := false
|
|
|
|
ctx, cancel := context.WithCancel(context.Background())
|
|
defer func() {
|
|
cancel()
|
|
conn.Close()
|
|
plannerUnsub()
|
|
wfUnsub()
|
|
infrastructure.ReleaseRefreshOwnership(ownedPeers, executionsID)
|
|
if !confirmed {
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
}
|
|
}()
|
|
|
|
pushCheck := realPushCheckfunc
|
|
if scheduled, err = pushCheck(ctx, conn, req, user, ws, executionsID, wfID, scheduled, asap, preemption, true); err != nil {
|
|
fmt.Println("UPDATE CONFIRM FIRST scheduled", err)
|
|
return
|
|
}
|
|
|
|
updateCh := make(chan infrastructure.WorkflowSchedule, 1)
|
|
closeCh := make(chan struct{})
|
|
go func() {
|
|
defer close(closeCh)
|
|
for {
|
|
var updated infrastructure.WorkflowSchedule
|
|
if err := conn.ReadJSON(&updated); err != nil {
|
|
return
|
|
}
|
|
select {
|
|
case updateCh <- updated:
|
|
default:
|
|
<-updateCh
|
|
updateCh <- updated
|
|
}
|
|
}
|
|
}()
|
|
|
|
for {
|
|
select {
|
|
case updated := <-updateCh:
|
|
fmt.Println("updated FOUND ", updated)
|
|
workflowScheduler := ws
|
|
schedulerMu.Lock()
|
|
if scheduler[user] != nil {
|
|
workflowScheduler = *scheduler[user]
|
|
}
|
|
schedulerMu.Unlock()
|
|
|
|
if updated.Confirm {
|
|
// Subscribe BEFORE calling Schedule to avoid missing the notification.
|
|
confirmCh, confirmUnsub := infrastructure.SubscribeSessionConfirmation(executionsID)
|
|
defer confirmUnsub()
|
|
|
|
workflowScheduler.UUID = executionsID
|
|
_, _, _, schedErr := infrastructure.Schedule(&workflowScheduler, wfID, req)
|
|
if schedErr != nil {
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
_ = conn.WriteJSON(map[string]interface{}{
|
|
"error": schedErr.Error(),
|
|
})
|
|
return
|
|
}
|
|
fmt.Println("UPDATE CONFIRM — waiting for execution confirmation")
|
|
select {
|
|
case <-confirmCh:
|
|
fmt.Println("UPDATE CONFIRM done")
|
|
confirmed = true
|
|
_ = conn.WriteJSON(map[string]interface{}{
|
|
"confirmed": true,
|
|
"scheduling_id": executionsID,
|
|
})
|
|
case <-time.After(60 * time.Second):
|
|
_ = conn.WriteJSON(map[string]interface{}{
|
|
"confirmed": false,
|
|
"error": "confirmation timeout: scheduling accepted but peers did not confirm in time",
|
|
})
|
|
case <-ctx.Done():
|
|
// client disconnected before confirmation
|
|
}
|
|
if !confirmed {
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
fmt.Println("UPDATE CONFIRM not done")
|
|
}
|
|
return
|
|
}
|
|
// Detect mode change before updating local vars.
|
|
modeChanged := (updated.Asap != nil && *updated.Asap != asap) ||
|
|
(updated.Preemption != nil && *updated.Preemption != preemption)
|
|
if updated.Asap != nil {
|
|
asap = *updated.Asap
|
|
}
|
|
if updated.Preemption != nil {
|
|
preemption = *updated.Preemption
|
|
}
|
|
changed := modeChanged ||
|
|
updated.Cron != workflowScheduler.Cron ||
|
|
!updated.Start.Equal(workflowScheduler.Start) ||
|
|
updated.DurationS != workflowScheduler.DurationS ||
|
|
(updated.End == nil) != (workflowScheduler.End == nil) ||
|
|
(updated.End != nil && workflowScheduler.End != nil && !updated.End.Equal(*workflowScheduler.End)) ||
|
|
updated.BookingMode != workflowScheduler.BookingMode ||
|
|
!reflect.DeepEqual(updated.SelectedBillingStrategy, workflowScheduler.SelectedBillingStrategy) ||
|
|
!reflect.DeepEqual(updated.SelectedInstances, workflowScheduler.SelectedInstances) ||
|
|
!reflect.DeepEqual(updated.SelectedPartnerships, workflowScheduler.SelectedPartnerships) ||
|
|
!reflect.DeepEqual(updated.SelectedBuyings, workflowScheduler.SelectedBuyings) ||
|
|
!reflect.DeepEqual(updated.SelectedStrategies, workflowScheduler.SelectedStrategies) ||
|
|
!reflect.DeepEqual(updated.SelectedEmbeddedStorages, workflowScheduler.SelectedEmbeddedStorages)
|
|
|
|
infrastructure.CleanupSession(executionsID, req)
|
|
|
|
schedulerMu.Lock()
|
|
scheduler[user] = &updated
|
|
schedulerMu.Unlock()
|
|
|
|
if scheduled, err = pushCheck(ctx, conn, req, user, ws, executionsID, wfID, scheduled, asap, preemption, changed || !scheduled); err != nil {
|
|
fmt.Println("UPDATE SCHEDULERD", err)
|
|
return
|
|
}
|
|
|
|
case remotePeerID := <-plannerCh:
|
|
workflowScheduler := ws
|
|
schedulerMu.Lock()
|
|
if scheduler[user] != nil {
|
|
workflowScheduler = *scheduler[user]
|
|
}
|
|
schedulerMu.Unlock()
|
|
if remotePeerID == selfPeerID {
|
|
if scheduled {
|
|
continue
|
|
}
|
|
result, checkErr := workflowScheduler.Check(wfID, asap, preemption, req)
|
|
if checkErr == nil {
|
|
result.SchedulingID = executionsID
|
|
_ = conn.WriteJSON(result)
|
|
}
|
|
continue
|
|
}
|
|
if scheduled, err = pushCheck(ctx, conn, req, user, ws, executionsID, wfID, scheduled, asap, preemption, scheduled); err != nil {
|
|
fmt.Println("UPDATE SCHEDULERD PLAN", err)
|
|
return
|
|
}
|
|
|
|
case <-wfCh:
|
|
if newPeers, err := infrastructure.GetWorkflowPeerIDs(wfID, req); err == nil {
|
|
plannerUnsub()
|
|
watchedPeers = newPeers
|
|
plannerCh, plannerUnsub = infrastructure.SubscribePlannerUpdates(newPeers)
|
|
newOwned := infrastructure.RequestPlannerRefresh(newPeers, executionsID)
|
|
ownedPeers = append(ownedPeers, newOwned...)
|
|
}
|
|
if scheduled, err = pushCheck(ctx, conn, req, user, ws, executionsID, wfID, scheduled, asap, preemption, false); err != nil {
|
|
fmt.Println("UPDATE WORKFLOW", err)
|
|
return
|
|
}
|
|
|
|
case <-closeCh:
|
|
fmt.Println("UPDATE Close ? ")
|
|
return
|
|
}
|
|
}
|
|
}
|
|
|
|
// @Title UnSchedule
|
|
// @Description unschedule a workflow execution: deletes its bookings on all peers then deletes the execution.
|
|
// @Param id path string true "execution id"
|
|
// @Success 200 {object} map[string]interface{}
|
|
// @router /:id [delete]
|
|
func (o *WorkflowSchedulerController) UnSchedule() {
|
|
user, peerID, groups := oclib.ExtractTokenInfo(*o.Ctx.Request)
|
|
executionID := o.Ctx.Input.Param(":id")
|
|
req := &tools.APIRequest{
|
|
Username: user,
|
|
PeerID: peerID,
|
|
Groups: groups,
|
|
Admin: true,
|
|
}
|
|
if err := infrastructure.UnscheduleExecution(executionID, req); err != nil {
|
|
o.Data["json"] = map[string]interface{}{"code": 404, "error": err.Error()}
|
|
} else {
|
|
o.Data["json"] = map[string]interface{}{"code": 200, "error": ""}
|
|
}
|
|
o.ServeJSON()
|
|
}
|
|
|
|
// @Title SearchScheduledDraftOrder
|
|
// @Description search draft order for a workflow
|
|
// @Param id path string true "id execution"
|
|
// @Param offset query string false
|
|
// @Param limit query string false
|
|
// @Success 200 {workspace} models.workspace
|
|
// @router /order/:id [get]
|
|
func (o *WorkflowSchedulerController) SearchScheduledDraftOrder() {
|
|
offset, _ := strconv.Atoi(o.Ctx.Input.Query("offset"))
|
|
limit, _ := strconv.Atoi(o.Ctx.Input.Query("limit"))
|
|
_, peerID, _ := oclib.ExtractTokenInfo(*o.Ctx.Request)
|
|
id := o.Ctx.Input.Param(":id")
|
|
filter := &dbs.Filters{
|
|
And: map[string][]dbs.Filter{
|
|
"workflow_id": {{Operator: dbs.EQUAL.String(), Value: id}},
|
|
"order_by": {{Operator: dbs.EQUAL.String(), Value: peerID}},
|
|
},
|
|
}
|
|
o.Data["json"] = oclib.NewRequestAdmin(orderCollection, nil).Search(filter, "", true, int64(offset), int64(limit))
|
|
o.ServeJSON()
|
|
}
|