It looks like everything is working as normal. Unless someone out there is getting a better response time on their set up. The heartbeat is continuous. Then when it does not get a response from the Master then the Backup takes over the role. Here is some information on Master/Backup relationship on the 55xx series switch.
Switching from the Master to the Master Backup
The Master Backup replaces the Stack Master if one or more of the following
events occur:
• The Stack Master fails or is removed from the stack.
• Links from the Stack Master to the stacking members fails.
• User performs soft switchover via the Web interface or the CLI.
Switching between the Stack Master and the Master Backup results in limited
service loss. Dynamic tables are relearned if a failure occurs. The Running
Configuration file is synchronized between Stack Master and the Master
Backup, and continues running on the Master Backup.
When a Master fails (or when an Operator requests a switchover between the Master and its Backup), a “switchover” occurs, the Backup master becomes a Master, it restarts all protocols and reconfigures the stack with the configuration it has. There is almost no abruption to the traffic within the stack except that no packets are sent from the stack members to the new stack master until it signals to the stack members that it is ready, thus minimizing the packet loss in case of Master failure.
As part of the takeover process, the “new” master erases all tables from the packet processors. These tables are relearned (Dynamic MAC addresses, STP state table).
Okay, well at least this means this behavior is expected.
If I read this correctly……. Hosts and VM’s on SWITCH2/BACKUP will still talk, as they’re on that switch. They will, however, not talk to the router or the load balancers, etc, perhaps (as that leaves that switch). But intra-switch traffic is fine. I did NOT test that, as I was testing plugged into a PC6248 stack, cabled to this stack via 2x10GbE in a LAG and then pinging a device and internal IP of the PC5548 stack. So my HOST was pinging REMOTE HOST and REMOTE SWITCH and they both failed. But this is suggesting that REMOTE HOST and REMOTE SWITCH would talk with no incident. I should test this.
Sufficiently though, as it won't talk outside of the switch either way, I need to ensure that those I report to understand that even with two switches, I'm not 100% independent or safe from loss. 0:45-2:00 is short, but not zero, and that's what they need to know. I think it's acceptable.
DELL-Willy M
802 Posts
1
February 1st, 2012 14:00
It looks like everything is working as normal. Unless someone out there is getting a better response time on their set up. The heartbeat is continuous. Then when it does not get a response from the Master then the Backup takes over the role. Here is some information on Master/Backup relationship on the 55xx series switch.
Switching from the Master to the Master Backup
The Master Backup replaces the Stack Master if one or more of the following
events occur:
• The Stack Master fails or is removed from the stack.
• Links from the Stack Master to the stacking members fails.
• User performs soft switchover via the Web interface or the CLI.
Switching between the Stack Master and the Master Backup results in limited
service loss. Dynamic tables are relearned if a failure occurs. The Running
Configuration file is synchronized between Stack Master and the Master
Backup, and continues running on the Master Backup.
When a Master fails (or when an Operator requests a switchover between the Master and its Backup), a “switchover” occurs, the Backup master becomes a Master, it restarts all protocols and reconfigures the stack with the configuration it has. There is almost no abruption to the traffic within the stack except that no packets are sent from the stack members to the new stack master until it signals to the stack members that it is ready, thus minimizing the packet loss in case of Master failure.
As part of the takeover process, the “new” master erases all tables from the packet processors. These tables are relearned (Dynamic MAC addresses, STP state table).
NetWise
7 Posts
0
February 1st, 2012 15:00
Okay, well at least this means this behavior is expected.
If I read this correctly……. Hosts and VM’s on SWITCH2/BACKUP will still talk, as they’re on that switch. They will, however, not talk to the router or the load balancers, etc, perhaps (as that leaves that switch). But intra-switch traffic is fine. I did NOT test that, as I was testing plugged into a PC6248 stack, cabled to this stack via 2x10GbE in a LAG and then pinging a device and internal IP of the PC5548 stack. So my HOST was pinging REMOTE HOST and REMOTE SWITCH and they both failed. But this is suggesting that REMOTE HOST and REMOTE SWITCH would talk with no incident. I should test this.
Sufficiently though, as it won't talk outside of the switch either way, I need to ensure that those I report to understand that even with two switches, I'm not 100% independent or safe from loss. 0:45-2:00 is short, but not zero, and that's what they need to know. I think it's acceptable.