Netflix/chaosmonkey · error

deploySchedule: could not publish schedule: %v

Error message

deploySchedule: could not publish schedule: %v

What it means

After computing 'today' in the local timezone, deploySchedule persists the schedule via ss.Publish (the SchedStore, e.g. backed by Cloud Datastore or chaosmonkey-api). If Publish returns an error, the schedule was not saved and this wrapped error is returned; cron registration is skipped. The underlying cause follows the colon.

Source

Thrown at command/schedule.go:91

	}

	return nil
}

// deploySchedule publishes the schedule to chaosmonkey-api
// and registers the schedule with the local cron
func deploySchedule(s *schedule.Schedule, ss schedstore.SchedStore, cfg *config.Monkey) error {
	loc, err := cfg.Location()
	if err != nil {
		return fmt.Errorf("deploySchedule: could not retrieve local timezone: %v", err)
	}

	today := time.Now().In(loc)

	err = ss.Publish(today, s)

	if err != nil {
		return fmt.Errorf("deploySchedule: could not publish schedule: %v", err)
	}

	err = registerWithCron(s, cfg)
	return err
}

// registerWithCron registers the schedule of terminations with cron on the local machine
//
// Creates or overwrites the file specified by config.Chaos.CronPath()
func registerWithCron(s *schedule.Schedule, cfg *config.Monkey) error {
	crontab := s.Crontab(cfg.TermPath(), cfg.TermAccount())
	var perms os.FileMode = 0644 // -rw-r--r--
	log.Printf("Writing %s\n", cfg.CronPath())
	err := ioutil.WriteFile(cfg.CronPath(), crontab, perms)
	return err
}

View on GitHub (pinned to eaa28fb761)

Solutions

1. Read the wrapped cause to identify whether it is a network, auth, or storage error.
2. Verify the schedstore backend (Datastore/chaosmonkey-api) is up and credentials have write access to the schedule entities.
3. Check network connectivity from the monkey host to the storage backend and retry after transient failures.
4. Validate the generated schedule entries if the cause indicates a bad-entity/payload error.

// preflight: confirm the schedstore backend is writable before deploying
if err := schedStoreHealthy(ss); err != nil {
    return fmt.Errorf("schedstore unavailable: %v", err)
}
// schedStoreHealthy issues a lightweight read/write probe per your SchedStore implementation

if err := do(dep, appCfgGetter, schedStore, cfg, constrainer, apps); err != nil {
    if strings.Contains(err.Error(), "could not publish schedule") {
        log.Errorf("publish failed: %v", errors.Unwrap(err))
        // retry with backoff for transient backend errors
        for i := 0; i < 3; i++ {
            time.Sleep(time.Duration(1<<i) * time.Second)
            if err2 := do(dep, appCfgGetter, schedStore, cfg, constrainer, apps); err2 == nil {
                return nil
            }
        }
        return err
    }
    return err
}

When it happens

Trigger: deploySchedule (via the schedule command) when ss.Publish(today, s) fails: storage backend unreachable, authentication/permission failure, or the store rejects the schedule payload.

Common situations: Cloud Datastore/API outages, missing service-account permissions to write schedule entities, network partition from the monkey server, or corrupted/oversized schedule data.

Related errors

• failed to populate schedule: %v
• failed to deploy schedule: %v

AI-assisted analysis of Netflix/chaosmonkey@eaa28fb761 (2026-09-03). Data as JSON: /api/errors/29524745a512afe4. Report an issue: GitHub.