Version 1.2.0
Scheduled jobs
Run one-shot workloads on timezone-aware schedules, with bounded retries, execution history, logs, and remote management.
Declare a job
A scheduled job runs a one-shot command and records its exit status. Building a job records a version without starting a service. Declare both schedule and job; job workloads do not accept port, health, process, or deploy blocks.
name: nightly-report
type: job
schedule:
cron: "0 3 * * *"
timezone: UTC
missed: skip
job:
command: ["./nightly-report"]
timeout: 2m
retries: 1
concurrency:
policy: forbidSchedule and execution
Schedules accept five-field cron expressions and hourly, daily, weekly, monthly, and yearly descriptors. The timezone is an IANA name and defaults to UTC. Only the monitor runs scheduled occurrences; a manual run works without it.
Missed occurrences are skipped without catch-up. Retries run sequentially, timeout stops the process group, and the default forbid policy prevents overlapping runs of the same job. Output is captured per execution with secrets redacted.
Run and inspect jobs
phelix config lint
phelix build nightly-report
phelix job run nightly-report
phelix job history nightly-report --limit 5
phelix job logs nightly-reportA manual run uses the same timeout, retry, and concurrency settings. History records success, failure, timeout, cancellation, skipped occurrences, and interrupted runs. Use job run rather than service start, stop, or restart commands.
Remote management and recovery
The connected monitor supports remote job execution, history, logs, and lifecycle events. Execution records are durable and active records are not pruned. Restart recovery distinguishes a live manual process from an interrupted execution.