AutoSys autorep — Filtering 10K Jobs With -s FA

autorep (short for 'auto report') is the command you'll run dozens of times every day as an AutoSys operator. It's how you check job statuses, view job definitions, and generate reports. Mastering autorep patterns makes you dramatically faster at monitoring and troubleshooting.

But autorep can be dangerous. Run autorep -J % in an environment with 10,000 jobs and you'll flood your terminal with 10,000 lines of output. The failures you care about scroll past before you see them. Without -s FA (filter failures) and -d (detailed view), you'll miss critical information. And without -q backups, an accidental jil < delete_all.jil is catastrophic.

By the end you'll know every useful autorep flag, practical query patterns for finding failed jobs, how to back up job definitions, and how to avoid the terminal-clogging trap that hides failures.

How AutoSys autorep -s FA Filters 10K Jobs Without Crashing

The autorep command in AutoSys is the primary interface for querying job definitions and statuses from the AutoSys database. The -s FA flag filters the report to show only jobs with a status of 'FAILURE' — meaning jobs that have terminated abnormally. This is not a live poll; autorep reads from the Event Server's database, which stores the last known status for each job. For a system managing 10,000+ jobs, autorep -s FA returns only the subset of jobs that failed, typically 1-5% of total jobs, making it efficient for daily operations. The command queries the database using a WHERE clause on the status column, so performance depends on database indexing. In practice, autorep -s FA is used in monitoring scripts and dashboards to quickly identify failed jobs without parsing the entire job stream. It's the go-to command for operations teams during shift handovers to assess system health — a single command can surface 50 failed jobs out of 10,000 in under a second, provided the database is tuned.

Basic autorep usage

The most common use is checking the status of one or more jobs.

# Check a single job
autorep -J daily_report

# Check all jobs (% is the wildcard in AutoSys)
autorep -J %

# Check all jobs matching a pattern
autorep -J eod_%        /* all jobs starting with eod_ */
autorep -J %_extract    /* all jobs ending with _extract */
autorep -J etl_*_load   /* wildcard in middle */

# Check all jobs inside a specific box
autorep -J eod_box%

# Check machine status
autorep -M %            /* all machines */
autorep -M prod-server-01

The -q flag — getting JIL definitions

The -q flag outputs the job definition in JIL format. This is essential for backups, auditing, and creating job templates.

# Get JIL definition of one job
autorep -J daily_report -q

# Backup all jobs to a file
autorep -J % -q > /backup/all_jobs_$(date +%Y%m%d).jil

# Backup all jobs in a box
autorep -J eod_box% -q > /backup/eod_jobs_backup.jil

# Compare with previous backup (useful for change auditing)
diff /backup/eod_jobs_20260318.jil /backup/eod_jobs_20260319.jil

The -d flag — detailed job information

The -d flag shows all attributes of a job including its current runtime and any active overrides — much more detail than the default status view.

# Detailed view of one job
autorep -J daily_report -d

# Check prior runs (-run flag)
autorep -J daily_report -run 1    /* previous run */
autorep -J daily_report -run 2    /* run before that */

# Filter by status
autorep -J % -s FA    /* all FAILURE jobs */
autorep -J % -s RU    /* all RUNNING jobs */
autorep -J % -s PE    /* all PEND_MACH jobs */

Why `-L` Flagging Is the Only Way to Audit Historical Runs

Your ops director wants to know why a job reran three times last Thursday. autorep -d shows you current status. Useless. You need the run history. That's the -L flag. It dumps the job's log file path, start/end times, and exit codes for every execution in the retention window.

Run autorep -L JOB_WEBHOOK_INGEST -r and you get the last N runs. No GUI. No database query. Five seconds. The -r flag filters to recent runs only — critical when your log retention is 30 days and the job runs hourly. Without it, you're reading 720 entries.

The output isn't pretty. It's raw, tabular, and exactly what a debugger needs. Exit code 12 across three consecutive runs? That's a file lock. Exit code 127 on the fourth? Missing binary. You can trace failure cascades without opening a single log file.

Senior shortcut: alias this in your .bashrc as jobhist and force your juniors to use it before escalating to you.

// io.thecodeforge — devops tutorial

// Get last 5 runs for a critical ETL job
autorep -L INVOICE_PAYLOAD_EXTRACT -r -5

// Output:
Job Name: INVOICE_PAYLOAD_EXTRACT
Job Type: CMD

#Run Start Time         End Time           Exit   Status
----- ------------------ ------------------ ------ --------
1     2024-12-12 02:00:01 2024-12-12 02:14:45 00000 SUCCESS
2     2024-12-12 02:15:00 2024-12-12 02:15:03 00001 FAILURE
3     2024-12-12 02:16:00 2024-12-12 02:16:02 00001 FAILURE
4     2024-12-12 02:17:00 2024-12-12 02:20:12 00000 SUCCESS

The `-w` Flag: Why You Don't Play Whack-a-Mole With Alerts

You've got 200 jobs. Three are on hold. Two are running late. One failed at midnight. autorep -w gives you a single line per job with status alone. No headers. No frame. Just the state machine.

Filter it with grep -E 'FA|TE|SU' and you see failures, terminations, and successes in one glance. No scrolling. No grep through a wall of JIL.

The real win? You can put this in a cron job. Every 5 minutes, run autorep -w JOB_GROUP_PROD | mail -s 'Autosys Pulse' ops-team@corp.com. That's your zero-config alerting. No Splunk. No PagerDuty. Just text.

Downside: -w doesn't show exit codes. You still need -d for that. But for a heartbeat check? It's the fastest way to know if your pipeline is haemorrhaging.

Production trick: Combine -w with -L in a shell loop. First -w finds the failing jobs. Second -L gets the logs. That's two commands to triage anything.

// io.thecodeforge — devops tutorial

// Check all jobs in PROD_BATCH group
// Only show FAILURE, TERMINATED, or RUNNING
autorep -w PAYMENTS_GROUP | grep -E 'FA|TE|RU'

// Output:
INVOICE_GENERATOR    RU  12/12 02:14:22
PAYMENT_VALIDATOR    FA  12/12 01:58:00
AUDIT_EXPORT         TE  12/12 01:30:00
SETTLEMENT_PUSH      SU  12/12 02:15:01