After the Enterprise Distributed Application Service (EDAS) Agent process stops on an Elastic Compute Service (ECS) instance, the Status column in the instance list displays Agent Exception. You cannot restart the ECS instance, or start, stop, or restart the application on the affected instance.
ECS instances without deployed applications are not affected by this issue.
Common causes
Before you restart the agent, review common causes to prevent the issue from recurring:
| Cause | How to check |
|---|---|
| Out-of-memory (OOM) killer terminated the process | Run free -h to check available memory |
| Agent process crashed | Check /home/admin/edas-agent/logs/agent.log for stack traces |
| Scheduled restart task missing or misconfigured | Run crontab -l to verify the agent restart cron job exists |
Restart the EDAS Agent
A cron job is configured to restart the EDAS Agent automatically. In most cases, manual intervention is not required. To verify, run crontab -l.
If the agent has not recovered automatically, restart it manually:
Log on to the ECS instance.
Run the following command to start the EDAS Agent:
sudo -u admin /home/admin/edas-agent/bin/startup.shGo back to the EDAS console and verify that the Status column for the instance no longer displays Agent Exception.
Troubleshoot persistent Agent Exception
If the Agent Exception status persists after you restart the agent, investigate the root cause.
Check available memory
Insufficient system memory can trigger the Linux OOM killer, which terminates the EDAS Agent process.
Log on to the ECS instance and check memory usage:
free -hIf available memory is critically low, free up resources or upgrade the instance specifications.
For more information, see OOM killer triggered by low physical memory.
Check the EDAS Agent logs
Review the EDAS Agent log file for error details:
tail -n 100 /home/admin/edas-agent/logs/agent.logLook for error messages or stack traces that indicate why the agent process stopped. For example, OutOfMemoryError or OOM-related stack traces indicate insufficient JVM or system memory. In this case, free up memory or upgrade the instance specifications.