studio/flamenco
studio/flamenco
12
30
Fork 38
Code Issues 64 Pull Requests 5 Releases Wiki Activity

Support pausing jobs #104313

Manually merged
Sybren A. Stüvel merged 28 commits from David-Zhang-10/flamenco:paused-job-status into main 2024-07-01 17:53:44 +02:00
Conversation 1 Commits 28 Files Changed 12 +11 -12
1 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 050c779995 - Show all commits

23
internal/manager/task_state_machine/task_state_machine.go
View File

@ -5,8 +5,6 @@ package task_state_machine
import ( import (
"context" "context"
"fmt" "fmt"
"projects.blender.org/studio/flamenco/pkg/website"
David-Zhang-10 marked this conversation as resolved Outdated
Sybren A. Stüvel commented 2024-06-17 15:42:40 +02:00
Outdated
Review
Copy Link

Keep the newline between built-in packages and 3rd party packages.

Keep the newline between built-in packages and 3rd party packages.
"github.com/rs/zerolog" "github.com/rs/zerolog"
"github.com/rs/zerolog/log" "github.com/rs/zerolog/log"
@ -120,6 +118,7 @@ func (sm *StateMachine) updateJobAfterTaskStatusChange(
return sm.jobStatusIfAThenB(ctx, logger, job, api.JobStatusCompleted, api.JobStatusRequeueing, "task was queued") return sm.jobStatusIfAThenB(ctx, logger, job, api.JobStatusCompleted, api.JobStatusRequeueing, "task was queued")
case api.TaskStatusPaused: case api.TaskStatusPaused:
// Pausing a task has no impact on the job.
return nil return nil
case api.TaskStatusCanceled: case api.TaskStatusCanceled:
@ -214,10 +213,19 @@ func (sm *StateMachine) updateJobOnTaskStatusCompleted(ctx context.Context, logg
if err != nil { if err != nil {
return err return err
} }
numActive, _, err := sm.persist.CountTasksOfJobInStatus(ctx, job, api.TaskStatusActive)
if err != nil {
return err
}
if numComplete == numTotal { if numComplete == numTotal {
logger.Info().Msg("all tasks of job are completed, job is completed") logger.Info().Msg("all tasks of job are completed, job is completed")
return sm.JobStatusChange(ctx, job, api.JobStatusCompleted, "all tasks completed") return sm.JobStatusChange(ctx, job, api.JobStatusCompleted, "all tasks completed")
} }
if numActive == 0 && job.Status == api.JobStatusPauseRequested {
// there is no active task, and the job is in pause-requested status, so we can pause the job
logger.Info().Msg("all tasks of job are completed, job is paused")
David-Zhang-10 marked this conversation as resolved Outdated
Sybren A. Stüvel commented 2024-06-19 10:40:48 +02:00
Outdated
Review
Copy Link

I don't think this is the right structure. The if {} block now contains a copy of the code below it, which is error-prone. There should be one piece of code responsible for one thing. By duplicating this code, there is the chance that one of the copies goes out of sync with the other, introducing hard to reason about bugs.

Also, don't use else after a return, it doesn't mean anything.

Better to just use:

if job.Status == api.JobStatusPauseRequested {
		numActive, _, err := sm.persist.CountTasksOfJobInStatus(ctx, job, api.TaskStatusActive)
		if err != nil {
			return err
		}
		if numActive == 0 {
			// There is no active task, and the job is in pause-requested status, so we can pause the job.
			failLogger.Info().Msg("No more active tasks, job is paused")
			return sm.JobStatusChange(ctx, job, api.JobStatusPaused, "all tasks completed")
		}
}

It'll also make it easier to move the code into a function of its own. It's the same "single responsibility principle" as before: there should be one piece of code that checks whether pausing a job is complete.

I don't think this is the right structure. The `if {}` block now contains a copy of the code below it, which is error-prone. There should be one piece of code responsible for one thing. By duplicating this code, there is the chance that one of the copies goes out of sync with the other, introducing hard to reason about bugs. Also, don't use `else` after a `return`, it doesn't mean anything. Better to just use: ```go if job.Status == api.JobStatusPauseRequested { numActive, _, err := sm.persist.CountTasksOfJobInStatus(ctx, job, api.TaskStatusActive) if err != nil { return err } if numActive == 0 { // There is no active task, and the job is in pause-requested status, so we can pause the job. failLogger.Info().Msg("No more active tasks, job is paused") return sm.JobStatusChange(ctx, job, api.JobStatusPaused, "all tasks completed") } } ``` It'll also make it easier to move the code into a function of its own. It's the same "single responsibility principle" as before: there should be one piece of code that checks whether pausing a job is complete.
return sm.JobStatusChange(ctx, job, api.JobStatusPaused, "all tasks completed")
}
logger.Info(). logger.Info().
Int("taskNumTotal", numTotal). Int("taskNumTotal", numTotal).
Int("taskNumComplete", numComplete). Int("taskNumComplete", numComplete).
@ -453,7 +461,6 @@ func (sm *StateMachine) pauseTasks(
// Any task that might run in the future should get paused. // Any task that might run in the future should get paused.
// Active jobs should remain active until finished // Active jobs should remain active until finished
taskStatusesToPause := []api.TaskStatus{ taskStatusesToPause := []api.TaskStatus{
api.TaskStatusActive,
api.TaskStatusQueued, api.TaskStatusQueued,
api.TaskStatusCanceled, api.TaskStatusCanceled,
api.TaskStatusSoftFailed, api.TaskStatusSoftFailed,
@ -466,15 +473,7 @@ func (sm *StateMachine) pauseTasks(
return "", fmt.Errorf("pausing tasks of job %s: %w", job.UUID, err) return "", fmt.Errorf("pausing tasks of job %s: %w", job.UUID, err)
} }
// If pause was requested, it has now happened, so the job can transition. return api.JobStatusPauseRequested, nil
if job.Status == api.JobStatusPauseRequested {
logger.Info().Msg("all tasks of job paused, job can go to 'paused' status")
return api.JobStatusPaused, nil
}
// This could mean state transition entered a non-recoverable error state.
log.Warn().Str("jobStatus", string(job.Status)).Msgf("unexpected job status in StateMachine::pauseTasks(), please report this at %s", website.BugReportURL)
return "", fmt.Errorf("unexpected job status %q in StateMachine::pauseTasks()", job.Status)
} }
// requeueTasks re-queues all tasks of the job. // requeueTasks re-queues all tasks of the job.