Pearl1594 opened a new issue, #14253: URL: https://github.com/apache/cloudstack/issues/14253
### problem The management server identity (msid) is computed at MS startup by iterating NetworkInterface.getNetworkInterfaces(), reversing the list, and selecting the first non-virtual, non-loopback physical interface. This ordering is controlled by the OS/kernel and is not stable across reboots, kernel updates, or configuration changes. When the selected interface changes (e.g., from ens35 to ens34), the management server boots with a new msid while retaining the same service_ip. This causes: 1. A new row is inserted into cloud.mshost with the new msid and state Up. 2. The old row remains as Down with removed = NULL indefinitely - no automatic cleanup. 3. KVM agents remember the old msid and place the new one on the avoidMSHostList, refusing to reconnect. 4. Manual systemctl restart cloudstack-agent on each host is required. Similar to : https://github.com/apache/cloudstack/issues/10915 ### versions Observed on 4.22.1.0 - but dates back to when the clustered management server feature was introduced. ### The steps to reproduce the bug 1. 2. 3. ... ### What to do about it? - Derive msid from a stable identifier - e.g., persist a UUID on first boot (file-based or DB-backed), not from a volatile MAC address. - When registering, detect if another mshost row already has the same service_ip and mark it removed=1. - Provide a config knob (management.server.id) for operators who need full control over the identity -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
