AutoSys Installation: The 5 Checks Before You Deploy
- AUTOSERV is the most important environment variable — wrong value = wrong instance. No error. No warning.
- Put AUTOSERV in your shell prompt. Always visible. Never forgotten.
- Port 7520 must be open between Event Processor and every agent. Closed port = jobs stay PENDING.
- AUTOSERV is the instance identifier — everything depends on it. Wrong AUTOSERV = wrong database = wrong jobs.
- Port 7520 must be open between Event Processor and every agent. Closed port = jobs stay PENDING forever.
- 8GB RAM minimum for small installs, 16GB+ for production. Underprovisioning causes event processor lag.
- AUTOUSER ($AUTOUSER/out/event_demon.$AUTOSERV) is your first stop for debugging. Logs tell you everything.
- Post-install test: chk_auto_up -A (processor running), autoflags -a (server reachable), test job end-to-end.
- Silent killer: no dedicated service account. Mixing AutoSys with application accounts = permission hell.
Install Verification — 60-Second Checklist
Verify instance and AUTOSERV
echo $AUTOSERVautoflags -aEvent Processor not running
chk_auto_up -Atail -50 $AUTOUSER/out/event_demon.$AUTOSERVAgents not responding
autoping -m AGENT_HOSTtelnet AGENT_HOST 7520No jobs visible
autoflags -a | grep Instanceecho $AUTOSERVProduction Incident
Production Debug GuideThe 5 things to check when a fresh install doesn't behave
Your infrastructure team handles the AutoSys install. You just use it. That's the theory.
Reality: you'll SSH into the AutoSys server at 2 AM to debug why a job isn't running. And if you don't know where the logs live or which AUTOSERV you're in, you'll waste 30 minutes figuring out the basics.
This guide assumes the install is done. We focus on the 5 checks that verify it's working — and the environment variables you'll type every single day.
Prerequisites before installation
Before AutoSys can be installed, you need:
On the AutoSys server: - A supported OS (RHEL/CentOS Linux, or Windows Server) - A supported database (historically Sybase ASE, now Oracle or other RDBMS) - Java runtime (AutoSys WCC runs on Tomcat) - Sufficient RAM — at least 8GB for a small installation, 16GB+ for production - A dedicated service account (commonly named autosys or appuser)
On each agent machine: - A supported OS (Linux, Windows, AIX, Solaris) - The AutoSys Agent package (lightweight, minimal footprint) - Network connectivity to the AutoSys server on the configured port
Port planning matters. The Event Processor communicates with agents over specific ports (default 7520). Make sure firewalls are open between the AutoSys server and every agent machine.
The silent killer: Using a shared account for multiple environments. If the same OS account connects to DEV and PRD, you will eventually run a command against the wrong instance.
# Check Java version (required for WCC) java -version # Check available memory free -h # Check disk space on AutoSys installation paths df -h /opt/CA /opt/autosys /u01 # Verify network connectivity to agent machines autoping -m agent-machine-01 # Check AutoSys port is open between server and agent nc -zv agent-machine-01 7520 # Verify separate service accounts (critical) whoami # Should be autosys_dev, autosys_qat, or autosys_prd — never 'autosys' shared
Total: 31Gi Used: 8.2Gi Free: 23Gi
Filesystem Size Used Avail Use%
/opt/CA 50G 12G 38G 24%
Connection to agent-machine-01 7520 port [tcp] succeeded!
Current user: autosys_dev
- Event Server = kitchen inventory system (database)
- Event Processor = head chef (reads orders, dispatches cooks)
- Remote Agent = wait station (each room has one)
- AUTOSERV = restaurant name — orders don't cross between restaurants
- Port 7520 = the pathway between kitchen and wait stations
Key environment variables you'll use every day
AutoSys uses several environment variables that control which instance you're talking to and where log files live. These are set in your shell profile (.bash_profile or .profile) for the AutoSys service account.
Knowing these by heart makes you much faster at debugging. When something goes wrong at 3 AM and you're SSH'd into the AutoSys server, these are the first things you check.
Most critical: AUTOSERV. It tells AutoSys which instance to connect to. Set it wrong, and you'll submit jobs to the wrong environment. No warning. No error. Just silent wrongness.
The 3 AM trick: If a job isn't behaving, echo $AUTOSERV and autoflags -a. If they don't match your expectation, fix the environment before touching anything else.
# The AutoSys instance name — critical, everything uses this export AUTOSERV=ACE # AutoSys installation directory export AUTOSYS=/opt/CA/WorkloadAutomationAE # AutoSys user/data directory (logs, config files, PID files) export AUTOUSER=/opt/CA/WorkloadAutomationAE/autouser # Path to AutoSys binaries (needed for jil, autorep, sendevent etc.) export PATH=$AUTOSYS/bin:$PATH # The "3 AM sanity check": put AUTOSERV in your prompt export PS1='[\u@\h:$AUTOSERV \W]\$ ' # Add the above to your .bash_profile for the autosys service account: # source this file or add these lines to ~/.bash_profile # Verify the environment is set correctly: echo "Instance: $AUTOSERV" echo "Install dir: $AUTOSYS" echo "User dir: $AUTOUSER" autoflags -a
Instance: ACE
Install dir: /opt/CA/WorkloadAutomationAE
User dir: /opt/CA/WorkloadAutomationAE/autouser
AutoSys r12.0 Build 213 — Event Processor: RUNNING
Verifying a fresh installation
After installation, run through this checklist before declaring the system ready:
- Event Processor is running —
chk_auto_up -A - Can connect to Event Server —
autoflags -a - Agents are registered and active —
autorep -M % - Can define a test job — create a simple JIL job
- WCC web interface is reachable — open the browser URL
- Test job runs successfully end-to-end
Test job best practice: Use a job that does something harmless but observable. Write to a test log file. Run it manually first on the agent machine to verify the command works. Then let AutoSys run it.
The rookie mistake: Assuming the test job works because the JIL insert succeeded. A successful insert only means the syntax was correct. It doesn't mean the command will run successfully on the agent.
# 1. Check Event Processor chk_auto_up -A # 2. Check overall system autoflags -a # 3. List all registered machines autorep -M % # 4. Insert and immediately force-run a test job jil << 'JIL' insert_job: INSTALL_TEST job_type: CMD command: echo 'AutoSys installation verified successfully' >> /tmp/autosys_test.log machine: autosys-server owner: autosys date_conditions: 0 JIL # Force start it sendevent -E FORCE_STARTJOB -J INSTALL_TEST # Wait a few seconds, then check its status autorep -J INSTALL_TEST # Check the test log cat /tmp/autosys_test.log # Verify the test job runs on the correct machine # The machine attribute must match the agent hostname # Clean up echo 'delete_job: INSTALL_TEST' | jil
MACHINE NAME STATUS MAX LOAD
autosys-server ACTIVE 100
agent-machine-01 ACTIVE 100
agent-machine-02 ACTIVE 100
Job Name Last Start Status
INSTALL_TEST 03/19/2026 14:22:01 SU
/tmp/autosys_test.log:
AutoSys installation verified successfully
| Environment Variable | What it controls | Example value | What breaks if wrong |
|---|---|---|---|
| AUTOSERV | AutoSys instance name (connects to correct database) | ACE, PRD, DEV | Wrong instance → wrong jobs. No error. Silent failure. |
| AUTOSYS | Installation directory path | /opt/CA/WorkloadAutomationAE | Commands fail with 'command not found' |
| AUTOUSER | User data and log directory | /opt/CA/WorkloadAutomationAE/autouser | Logs go to wrong place; event_demon can't write |
| PATH | Must include AutoSys bin directory | $AUTOSYS/bin:$PATH | sendevent, autorep, jil not found |
🎯 Key Takeaways
- AUTOSERV is the most important environment variable — wrong value = wrong instance. No error. No warning.
- Put AUTOSERV in your shell prompt. Always visible. Never forgotten.
- Port 7520 must be open between Event Processor and every agent. Closed port = jobs stay PENDING.
- Test job on local server proves nothing about remote agents. Test EACH agent machine.
- 8GB RAM minimum, 16GB+ for production. Underprovisioning causes silent performance collapse.
- First command after SSH: autoflags -a. Verify BEFORE any destructive operation.
⚠ Common Mistakes to Avoid
Interview Questions on This Topic
- QWhat environment variables does AutoSys use and what does each control?JuniorReveal
- QWhat is the AUTOSERV variable in AutoSys?JuniorReveal
- QWhat port does the AutoSys Event Processor use to communicate with agents?JuniorReveal
- QHow do you verify an AutoSys installation is working correctly?Mid-levelReveal
- QWhat would you check first if AutoSys was freshly installed but jobs aren't running?Mid-levelReveal
- QHow do you safely manage multiple AutoSys instances (DEV, QAT, PRD) on the same server?SeniorReveal
Frequently Asked Questions
What is AUTOSERV in AutoSys?
AUTOSERV is an environment variable that identifies the AutoSys instance name. It's used by all AutoSys commands to know which database instance to connect to. If you have DEV, QAT, and PRD instances, each will have a different AUTOSERV value.
Critical safety issue: There is no safety prompt. If AUTOSERV is wrong, you'll run commands against the wrong environment. Always verify with autoflags -a before destructive operations.
What port does AutoSys use?
The default port for AutoSys agent communication is 7520. This port must be open in firewalls between the AutoSys Event Processor (server) and every agent machine. If the port is blocked, agents show as INACTIVE and jobs stay PENDING forever.
The WCC web interface uses port 8080 (or 8443 for HTTPS) by default.
How do I check if AutoSys is running?
chk_auto_up -Achecks Event Processor status (RUNNING/STOPPED)autoflags -achecks overall system status including Event Server connection
Both commands are available from the command line on the AutoSys server after sourcing the AutoSys environment variables.
Can AutoSys run on Linux?
Yes. AutoSys fully supports RHEL and CentOS Linux for both the server and agent components. It also supports Windows Server, AIX, and Solaris for agent machines. The Event Processor and Event Server typically run on Linux in production environments.
How many AutoSys instances can run on one server?
Multiple instances can run on the same physical or virtual server. Each instance has its own AUTOSERV identifier and its own database (Event Server). This is commonly used to have separate DEV, QAT, and PRD AutoSys environments on the same hardware.
Safety requirement: Use separate OS service accounts for each instance (autosys_dev, autosys_prd). Shared accounts inevitably lead to 'oops, wrong instance' incidents.
What's the first thing I should do after SSH'ing into an AutoSys server?
Run autoflags -a and verify AUTOSERV. This tells you exactly which instance you're connected to. Then check chk_auto_up -A to confirm the Event Processor is running.
This 5-second check prevents the most common mistake: thinking you're in DEV when you're actually in PRD. Put AUTOSERV in your prompt so you never forget.
Developer and founder of TheCodeForge. I built this site because I was tired of tutorials that explain what to type without explaining why it works. Every article here is written to make concepts actually click.