D-Link DGS-3100 user manual Replacing a Failed Stack Master

DGS-3100 Series Gigabit Stackable Managed Switch User Manual

–If the new stacking member has a Unit ID of 1 or 2, i.e. the stacking member is a master enabled unit, the new stacking member initiates the Master Election process. However, since the running Stack Master has a longer up-time, it remains the Stack Master and the new stacking member does not become a new Stack Master.

–If the new stacking member has a Unit ID of 3 to 6, the new unit attempts to become a stacking member, and is subject to the already running Stack Master. The Master Election process does not occur.

•The Stack Master performs a Unit ID Allocation and Conflict Resolution process.

–If the new stacking member was in Factory Default mode (the unit does not have an assigned Unit ID). The new stacking member is assigned the lowest available Unit ID by the Stack Master. It is strongly recommended that automatic assigned Unit ID mode be used since it provides improved stack resiliency.

–If the new stacking member already has an assigned Unit ID, and that Unit ID is unused in the current stack, the new stacking member retains its assigned Unit ID. The Stack Master applies any configuration relevant to that Unit ID.

–If the new stacking member already has an assigned Unit ID, and that Unit ID conflicts with an existing Unit ID the Stack Master reallocates a new Unit ID to the new stacking member. , The lowest available Unit ID is applied to the new stacking member. This occurs only if new stacking member does not have a manually assigned Unit ID, which the Stack Master cannot change.

–If the new stacking member cannot be assigned an available Unit ID, then the new stacking member is effectively shut down and cannot join the stack. For example, the stacking member replacement can only occur if the new stacking member has a manually assigned Unit ID.

•The Stack Master now performs Unit and Stacking Port Configuration for the new stacking member.

•Any configuration information which the Stack Master stores that is relevant to the Unit ID is assigned to the new stacking member. If the new stacking member was assigned the same Unit ID of the replaced stacking member, then the new stacking member receives the same configuration as the failed stacking member. If the new stacking member is identical to the replaced stacking member, the entire configuration of the replaced stacking member is applied to the new stacking member. The stack reverts to the stacking state it was running in before stacking member failed. However, sometimes the new stacking member is not identical to the failed stacking member. The Stack Master applies the configuration as follows:

•If a 24-port switch replaces a failed 48-port switch, then the new stacking member’s ports are configured according first 24 ports configuration of the failed stacking member.

NOTE: The 48 port configuration of the failed stacking member is recorded, even though only the first 24 port configuration is currently applied. If a 48 port switch is inserted and assigned the same Unit ID, the switch is configured with the port configuration of the original 48 port stacking member.

•If a 48-port stacking member replaces a 24-port stacking member, then the first 24 ports of thenew stacking member are configured according failed stacking member’s ports configuration. The remaining new stacking member ports are configured with the switch’s default settings.

Replacing a Failed Stack Master

This example assumes that that a stacking member acting as the Stack Master has failed in a running stack. When the system administrator is notified of the Stack Master failure and replaces existing Stack Master with a new switch.

When the Stack Master fails, the stack’s Backup Master recognizes the failure and that the Stack Master no longer responds and assumes the role of Stack Master. The Backup Master uses Master Discovery process to identify the failure. In case of Ring topology the Backup Master directs all other stack members to route traffic around the failed stacking member. At the same time the Backup Master notifies the system administrator of the failure using SYSLOG messages and SNMP traps.

When the failed stacking member is disconnected from the stack, all traffic is already routed around the failed stacking member. If all other stacking connections are left intact, the stack keeps running. When a new stacking member is inserted and powered up, the following occurs:

•The new stacking member performs Master Discovery process, and participates in a Master Election process. For more information on the Master Election, see Electing a Stacking Master.

–If the new stacking member has a Unit ID of 1 or 2, i.e. the stacking member is a master enabled; Master Election process is initiated. Since the running stack Backup Master has a longer run timeand if the Backup Master has been running for more then 10 minutes, the Backup Master remains the elected Stack Master. The new stacking member does not become the new Stack Master. This may result in new stacking member using

21