How to restart Winbind service with minimal service disruption
Applies to
- SGRID 9.x
- Both Windows and Linux
Description
It may be desirable to restart the Winbind services of an Active Gateway Node with minimal service disruption. For example, the Winbind service may leak memory and a restart would remedy the immediate concern.
The Winbind service is managed by the FSG heartbeat system. The FSG will consider an error to have occurred if a managed service restarts without the FSGs involvement. If the node is a member of a High Availability (HA) Gateway Node cluster (HAGC), restarting a managed service will trigger a failover to the Standby Gateway Node of the cluster.
It would be best to perform this when there is low customer activity on the Active Primary FSG. Please wait until queue sizes, such as "Files Stored to Grid: Pending" and "Files Retrieved from Grid: Pending" are near 0. A failover will typically take 10 seconds if there is no queue on the FSG cluster.
This procedure is tested against all current StorageGRID versions, however, you are encouraged to practice it in the lab prior to performing it on a production system.