Troubleshooting, Diagnostics, and Automation
Domain 9 asks you to diagnose why a device is not polled or an endpoint fails authentication, and then automate the fix: the diagnostics toolbox, log locations and the workflow engine.
Diagnostics toolbox
- Per-device tools include ping, traceroute, SNMP walk and an SSH/Telnet session launched from the server using stored CLI credentials - the fastest proof of whether routing, credentials or an ACL is at fault.
- Poll status and last-successful-poll separate unreachable devices from reachable-but-unmanaged ones. A credential test validates a username and secret before you apply it broadly.
- Endpoint/MAC search maps a MAC address to switch and port. Control diagnostics show which authentication and authorisation rule matched, and test-authentication replays a MAC through the rules.
- Server side, Administration > Diagnostics > Server Logs offers adjustable log levels, and a support-information bundle collects logs and state for Extreme support.
Automation
- Workflows link a trigger - alarm, event, schedule, endpoint authentication, discovery - to activities such as send email, run a CLI script, deploy configuration, apply a role, add to a site, or disable a switch port.
- Scripts run with device credentials from the credential list, not the admin's login, and use per-family command syntax. Failed activities stop the branch unless error handling is configured.
- Use scheduled tasks for recurring reports, backups and compliance checks, and the REST API for inventory and Control data when integrating with ticketing or a CMDB.
Worked example
A port must shut down when a rogue-device alarm fires. First build the alarm definition matching the rogue-detection trap, then a workflow with an alarm trigger filtered to that exact alarm name and an action that runs a script on the reporting device: enter interface context, issue shutdown, save. Test with a lab alarm - if the workflow never starts, the trigger filter name usually does not match the alarm exactly.
Traps that cost marks
- Testing credentials with the admin account instead of the configured device credential.
- Alarm-name mismatch between definition and workflow filter.
- Scripts assuming one device family's syntax for another family.
- Expecting alarms for data the device never sends.
- Capturing a support bundle after the log window has rolled over.