OOBM and
active
and
inactive
fragments
When there are two fragments are the same size and OOBM is enabled and configured for all members of the stack, then the fragment that contains the Commander will be
Active
and the other fragment becomes
Inactive
. However, if OOBM is
not enabled and configured for all members of the stack, then each fragment will be
Active
. To resolve this issue, OOBM must be enabled and configured.
Below are the behaviors observed with two-member stack in a chain configuration and the stacking cable is removed to simulate a failure that results in two identical size fragments:
switch# show stack Stack ID : 01002c59-e51c5bc0 MAC Address : 2c59e5-1c5be3 Stack Topology : Chain Stack Status : Active Split Policy : One-Fragment-Up Uptime : 0d 0h 13m Software Version : WB.15.16.0005 Mbr ID Mac Address Model Pri Status --- ------------- -------------------------------------- --- --------------- 1 2c59e5-1c5bc0 HPE J9726A 2920-24G Switch 128 Commander 2 d4c9ef-a49e00 HPE J9726A 2920-24G Switch 128 Standby
switch# show stack stack-ports Member Stacking Port State Peer Member Peer Port ------ ------------- -------- ----------- --------- 1 1 Up 2 2 1 2 Down 0 0 2 1 Down 0 0 2 2 Up 1 1
Results without OOBM configured:
From physical console of Member 1 (original stack commander):
switch# show stack Stack ID : 01002c59-e51c5bc0 MAC Address : 2c59e5-1c5be3 Stack Topology : Chain Stack Status : Fragment Active Split Policy : One-Fragment-Up Uptime : 0d 0h 09m Software Version : WB.15.16.0005 Mbr ID Mac Address Model Pri Status --- ------------- -------------------------------------- --- --------------- 1 2c59e5-1c5bc0 HPE J9726A 2920-24G Switch 128 Commander 2 d4c9ef-a49e00 HPE J9726A 2920-24G Switch 128 Missing
From physical console of Member 2:
switch# show stack Stack ID : 01002c59-e51c5bc0 MAC Address : 2c59e5-1c5be3 Stack Topology : Chain Stack Status : Fragment Active Split Policy : One-Fragment-Up Uptime : 0d 0h 9m Software Version : WB.15.16.0005 Mbr ID Mac Address Model Pri Status --- ------------- -------------------------------------- --- --------------- 1 2c59e5-1c5bc0 HPE J9726A 2920-24G Switch 128 Missing 2 d4c9ef-a49e00 HPE J9726A 2920-24G Switch 128 Commander
Note that both switches are Commanders and both Fragments remain
Active
.
Results with OOBM configured:
OOBM is now enabled with both oobm ports connected to the same network segment:
switch# sh oobm ip IPv4 Status : Enabled IPv4 Default Gateway : 10.238.255.254 | Address Interface Member | IP Config IP Address/Prefix Length Status Status ------ + --------- ------------------------- -------- --------- Global | manual 10.238.123.37/16 Active Up 1 | manual 10.238.123.37/16 Active Up 2 | manual 10.238.123.39/16 Active Up
switch# show stack (from console of member 1) Stack ID : 01002c59-e51c5bc0 MAC Address : 2c59e5-1c5be3 Stack Topology : Chain Stack Status : Fragment Active Split Policy : One-Fragment-Up Uptime : 0d 0h 16m Software Version : WB.15.16.0005 Mbr ID Mac Address Model Pri Status --- ------------- -------------------------------------- --- --------------- 1 2c59e5-1c5bc0 HPE J9726A 2920-24G Switch 128 Commander 2 d4c9ef-a49e00 HPE J9726A 2920-24G Switch 128 Missing
switch# show stack (from console of member 2) Stack ID : 01002c59-e51c5bc0 MAC Address : 2c59e5-1c5be3 Stack Topology : Chain Stack Status : Fragment Inactive Split Policy : One-Fragment-Up Uptime : 497d 1h 57m Software Version : WB.15.16.0005 Mbr ID Mac Address Model Pri Status --- ------------- -------------------------------------- --- --------------- 1 2c59e5-1c5bc0 HPE J9726A 2920-24G Switch 128 Missing 2 d4c9ef-a49e00 HPE J9726A 2920-24G Switch 128 Commander
Note that both switches are Commanders, but that the stack fragment containing member 2 is listed as "Fragment Inactive
". This is the correct behavior and why it is important that OOBM is enabled and configured for all members of the stack.