Monitoring Scheduled Jobs

Scheduled jobs include cron jobs, systemd timers, and any command that runs on a predictable schedule. These jobs often fail silently—the script errors out, but nothing alerts you until downstream systems break.

Configuration

Set your API key as an environment variable. For cron jobs, add it to your crontab:

# Edit crontab
crontab -e

# Add at top of crontab
PAKYAS_API_KEY=pk_live_xxxxx

# Then your job
0 2 * * * pakyas monitor nightly-job -- /path/to/script.sh

Or source from a file:

0 2 * * * . ~/.pakyas_env && pakyas monitor nightly-job -- /path/to/script.sh

See Environment Variables for all options.

When to use this

• Jobs run on a schedule (hourly, daily, weekly)
• Missing a run is a failure
• Silent failures are possible (script exits non-zero but no one notices)

Basic example

pakyas monitor nightly-job -- ./script.sh

Pakyas wraps your command, sends a start signal, captures output, and reports success or failure automatically.

Scheduler setup

# crontab example - runs every night at 2am
0 2 * * * pakyas monitor nightly-job -- /path/to/script.sh

What Pakyas detects

• Job exits non-zero
• Job runs longer than expected
• Job never reports back (missed schedule)

How schedules are evaluated

Pakyas supports two schedule types, each with different behavior for computing when the next ping is expected.

Schedule Type	Next Expected Based On	Best For
Interval	Last success/fail + period	Heartbeat monitoring
Cron	Next cron tick (wall-clock)	Fixed-time jobs

Interval schedules

Timer resets after each ping. If you set a 5-minute interval and your job finishes at 12:07, the next expected ping is 12:12.

Behavior:

• Next expected = last completion time + interval
• Long-running jobs “shift” the schedule forward
• Best for heartbeat-style monitoring where you care about time between check-ins

Example: A health check that should run every 5 minutes. If it takes 2 minutes to run, the next check starts 5 minutes after the previous one finished.

Cron schedules

Expected at fixed wall-clock times. If your cron is */5 * * * * (every 5 minutes), Pakyas expects pings at :00, :05, :10, etc.—regardless of how long the previous job ran.

Behavior:

• Next expected = next scheduled time from cron expression
• Long-running jobs become “Overrunning” instead of shifting the schedule
• Best for jobs that must run at specific times (backups at midnight, reports at 9am)

Example: A backup job scheduled at 0 2 * * * (2am daily). If it takes 3 hours, the next expected is still 2am tomorrow—not 5am.

Overrunning detection

For cron schedules, if a job is still running when the next scheduled time passes, Pakyas marks it as Overrunning. This is different from “Late” (which means no job started):

• Late/Missing: Expected time passed and no ping was received (job didn’t start)
• Overrunning: Job started but is still running past the next scheduled time