AutoSys sendevent Command — FORCE_STARTJOB Pitfalls

# Basic syntax
sendevent -E EVENT_TYPE -J job_name

# For a specific AutoSys instance
sendevent -E EVENT_TYPE -J job_name -S INSTANCE_NAME

# ── The most common sendevent operations ──────────────────────────

# Force start a job (run immediately regardless of conditions)
sendevent -E FORCE_STARTJOB -J daily_report

# Start a job normally (only if conditions are met)
sendevent -E STARTJOB -J daily_report

# Kill a running job
sendevent -E KILLJOB -J daily_report

# Put a job on hold (won't run until released)
sendevent -E JOB_ON_HOLD -J daily_report

# Release a job from hold
sendevent -E JOB_OFF_HOLD -J daily_report

# Put a job on ice (stronger suspension)
sendevent -E JOB_ON_ICE -J daily_report

# Release from ice
sendevent -E JOB_OFF_ICE -J daily_report

# Delete a job via event
sendevent -E DELETEJOB -J old_legacy_job

# Manually change a job's status
sendevent -E CHANGE_STATUS -J job_name -s SUCCESS

# Set a global variable
sendevent -E SET_GLOBAL -G "TRADE_COUNT=5432"

# Stop the Event Processor gracefully
sendevent -E STOP_DEMON

# Manually mark a job as SUCCESS (to unblock downstream dependencies)
sendevent -E CHANGE_STATUS -J extract_job -s SUCCESS

# Manually mark a job as FAILURE (to block downstream from starting)
sendevent -E CHANGE_STATUS -J questionable_job -s FAILURE

# After CHANGE_STATUS to SUCCESS, downstream jobs that were waiting
# will now evaluate their conditions and may start automatically

#!/bin/bash
# Auto-restart all FAILURE jobs matching a pattern

PATTERN="PRD_TRADING_%"

echo "Checking for failed jobs matching: $PATTERN"
FAILED_JOBS=$(autorep -J $PATTERN -s FA | awk 'NR>2 {print $1}' | grep -v '^$')

if [ -z "$FAILED_JOBS" ]; then
    echo "No failed jobs found"
    exit 0
fi

for JOB in $FAILED_JOBS; do
    echo "Restarting: $JOB"
    sendevent -E RESTART -J "$JOB"
    sleep 2   # brief pause between events to avoid overwhelming Event Processor
done

echo "Restart events sent for: $(echo $FAILED_JOBS | wc -w) jobs"

# Pattern 2: Set a global, then trigger downstream job
RECORD_COUNT=$(wc -l < /data/processed.csv)
sendevent -E SET_GLOBAL -G "TRADING_COUNT=${RECORD_COUNT}"
sendevent -E STARTJOB -J PRD_TRADING_VALIDATE_COUNT

# RESTART - only works if job status is FAILURE or has next cycle pending
sendevent -E RESTART -J export_daily   # re-runs with original conditions

# DELETEJOB - removes job definition permanently; use with extreme care
sendevent -E DELETEJOB -J old_job

# SET_GLOBAL - sets a global variable visible to all jobs
sendevent -E SET_GLOBAL -G "ENVIRONMENT=STAGING"

# SENDSTATUS - sends an application status (rarely used, but available)
sendevent -E SENDSTATUS -J my_job -s "Data validated OK"

# STOP_DEMON - graceful shutdown of Event Processor
sendevent -E STOP_DEMON

# START_DEMON - restart Event Processor (usually via script, not sendevent)
# NOTE: There is no START_DEMON event; use the autostart mechanism

# Step 1: Is the Event Processor running?
autorep -W   # returns NOSUCH? If yes, Event Processor is down

# Step 2: What is the job's current status?
autorep -q jobname | head -20

# Step 3: Check event log for recent events
tail -100 $AUTOUSER/events/events.log | grep jobname

# Step 4: Verify job name spelling and case
autorep -J "JOBNAME*" | grep -i jobname

# Step 5: Test connectivity to Event Processor host
ping autoevent_host

# Step 6: Check for multiple instances
sendevent -E PING   # sends a test event; responds with 'PONG' from Event Processor