02-08-2012 02:37 AM - edited 03-07-2019 04:48 AM
I have an existing stack of 4 x 2960-S switches connected by stack cables
I would like to add another 2960-S switch to the stack but am unable to as the 2960-S will only allow 4 x 2960-S switches per stack.
Can you advise how I would add the 5th 2960-S switch to the existing stack of 4 x 2960-S switches
Solved! Go to Solution.
02-28-2013 02:39 PM
OK, so here's how it works. Tested with 6 2960's. You CAN connect more than 4 switches via the stack cables while all switches are powered off, as the documentation suggests. Don't bother trying powered on, the 5th switch and on will stall at disabling stack ports and never recover. But when you power on all the switches, the master election takes place and three other switches are thrown in the stack. On all remaining switches, as the documentation indicated, become independant masters. The stack ports are automatically disabled. In this scenario, the independant masters would be considered provisioned because they have the master config on them. I was hoping there would be some redundancy with this config like if a switch failed it would automatically bring one of the independant masters into the stack to fill the void, but no such luck. The stack ports stay disabled until manually enabled again on the independant masters. Once you enable a stack port on an inependant master after one switch fails, it reboots and integrates into the stack. The only advantage I see here is that it eliminates having to plug in some stacking cables. The bad part about this design is that you have to reboot the switch to integrate it back into a stack and due to the way the stacking cables need to be cabled, you lose a redundant stacking connection on either end of the stack to a disabled stack port on an independant master. Theoretically you could configure a stack within the stack by keeping the independant stack masters separated with the disabled ports and enabling the connections between the independant masters, but I'm not sure I see any advantages there. So for my purposes, I think it's easier to stop at 4 (or less) and start a new stack at that point.
Hope this helped provide some insight on the topic. This is just my initial observations. I'll post more if I learn anything new.
On a side note, I'm not sure if others operate the way I do, but I come to these forums as a last resort. When someone just regurgitates some documentation to answer a question I find it insulting, because I've already read the documentation and now have questions that require further elaboration. So by doing so, you've wasted both your time and mine since I've had to read the same thing twice in order to sift and find a comprehensive answer. Just something to think about for future responses.
02-08-2012 02:44 AM
Hi,
Cisco FlexStack allows the stacking of up to four switches, with a throughput of 20 gigabits per second.
for more details refer below URL,
http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps6406/qa_c67-577519.html
Regards,
Vaibhav Shirkul
02-08-2012 02:46 AM
Hi Derrick,
The FlexyStack (2960) have only 4 units limit where the StackWise (3750) have 9 units limit.
You can not add more than 4 switches in FlexyStack.
For your information see the below table for difference between the above two stack technologies.
Cisco Catalyst 2960-S Stacking | Cisco Catalyst 3750-X StackWise | |
Device limit | 4 units | 9 units |
Stack bandwidth | 10G/20G | 32G/64G |
Architecture | Hardware drop table | Ring (destination stripping) |
Dynamic ring load balancing | No | Yes |
Stack convergence | 1-2 seconds | Few milliseconds |
Stack quality of service (QoS) | Applied hop by hop | Applied on ingress |
Management | Single IP address, SNMP, syslog | Single IP address, SNMP, syslog |
Configuration | Single config and CLI, auto image and config update | Single config and CLI, auto image and config update |
Show and debug commands | Unified | Unified |
Single forwarding and control plane | Synchronize Address Resolution Protocol (ARP), MAC address, Internet Group Management Protocol (IGMP), VLAN tables | Synchronize Address Resolution Protocol (ARP), MAC address, Internet Group Management Protocol (IGMP), VLAN tables, routing tables |
Cross-stack features | Yes | Yes |
Single bridge-ID | Yes | Yes |
Redundancy | Stack master 1:N redundancy | Stack master 1:N redundancy |
Hope the above clear and understand you.
Please rate all the helpfull posts.
Regards,
Naidu.
02-08-2012 02:59 AM
can you confirm the best practice for connecting a 5th 2960-S switch to the stack as I am required to connect another 2960-S switch to the existing stack of 4 x 2960-S switches
Should I just create a trunk on the 4th member of the stack and create a trunk on the 2960-S (5th switch) and connect the 5th switch to to the existing 4 x 2960-S stack via a SFP in both the 4th switch in the stack and the 2960-S (5th switch) run a fibre cable between both the 4th switch in the stack and the 2960-S (5th switch)
02-08-2012 03:26 AM
I dont understand why you are not able to understand the fact.
You can not add 5th switch to existing stack (2960 FlexyStack) this is a rule.
Hope the above clear and understand you.
Please rate all the helpfull posts.
Regards,
Naidu.
02-08-2012 03:53 AM
Hi Derrick,
You have two options here:
1. Connect the 5th 2960-s switch to the existing stack using the trunk links. You can configure these trunk links as cross-stack ether-channel ports and you will have the 5th switch connected to the stack via ether-channel working as an active active switch.
2. Break you existing stack to a stack of 2 switches each and then add the new 2960-s to eiather of the stack. You can connect these 2 stacks together via the ether-channel links and also conneect each stack to the upstream switch via dual links. You can run cross-stack ether-channel as well to the upstream switch depending how your present topology looks like.
Hope this helps.
Cheers,
-amit singh
02-08-2012 10:46 PM
can you confirm the best practice for connecting a 5th 2960-S switch to the stack as I am required to connect another 2960-S switch to the existing stack of 4 x 2960-S switches
No one will confirm this because you cannot and WILL NOT be able to stack a 5th 2960S using FlexStack.
You can create a new stack if you want.
02-28-2013 09:23 AM
I was curious how to do this too. I know some have said it's not possible, but this is directly from the Catalyst 2960 and 2960-S Software Configuration Guide under managing switch stacks:
It sure seems as though it's possible.
02-28-2013 01:27 PM
Cisco Catalyst 2960-S and 2960-SF Series Switches
Q. What is Cisco FlexStack stacking?
A. Cisco FlexStack is a hot-swappable, modular stacking solution that provides a true stacking capability, with all switches acting as a single switching unit with a unified data plane, using a single IP address. Cisco FlexStack is supported by the Catalyst 2960-S and 2960-SF series and allows the stacking of up to four switches, with a throughput of 20 gigabits per second.
02-28-2013 01:41 PM
It doesn't really do me any good to post more Cisco documentation since the official Cisco documentation conflicts with itself. I guess I'll have to rely on real world testing to answer this definitively. I'll post my findings.
02-28-2013 02:39 PM
OK, so here's how it works. Tested with 6 2960's. You CAN connect more than 4 switches via the stack cables while all switches are powered off, as the documentation suggests. Don't bother trying powered on, the 5th switch and on will stall at disabling stack ports and never recover. But when you power on all the switches, the master election takes place and three other switches are thrown in the stack. On all remaining switches, as the documentation indicated, become independant masters. The stack ports are automatically disabled. In this scenario, the independant masters would be considered provisioned because they have the master config on them. I was hoping there would be some redundancy with this config like if a switch failed it would automatically bring one of the independant masters into the stack to fill the void, but no such luck. The stack ports stay disabled until manually enabled again on the independant masters. Once you enable a stack port on an inependant master after one switch fails, it reboots and integrates into the stack. The only advantage I see here is that it eliminates having to plug in some stacking cables. The bad part about this design is that you have to reboot the switch to integrate it back into a stack and due to the way the stacking cables need to be cabled, you lose a redundant stacking connection on either end of the stack to a disabled stack port on an independant master. Theoretically you could configure a stack within the stack by keeping the independant stack masters separated with the disabled ports and enabling the connections between the independant masters, but I'm not sure I see any advantages there. So for my purposes, I think it's easier to stop at 4 (or less) and start a new stack at that point.
Hope this helped provide some insight on the topic. This is just my initial observations. I'll post more if I learn anything new.
On a side note, I'm not sure if others operate the way I do, but I come to these forums as a last resort. When someone just regurgitates some documentation to answer a question I find it insulting, because I've already read the documentation and now have questions that require further elaboration. So by doing so, you've wasted both your time and mine since I've had to read the same thing twice in order to sift and find a comprehensive answer. Just something to think about for future responses.
03-01-2013 02:03 AM
Hello Nathan,
you have been kind to provide a feedback on your tests with 6 2960-S.
However, the kind of answers you have received are to be considered normal and appropriate, because actually most of original posters/users usually need a reference to documentation or some guidance. It would be a risk to suggest something that is out of best practices to someone, that may not be experienced enough to manage a non standard setup.
>> You CAN connect more than 4 switches via the stack cables while all switches are powered off, as the documentation suggests.
The physical connection is only one part of the process of building a stack. A protocol is spoken by all stack members and candidate members on the stack ports.
Your tests show that the limitation of 4 switches is a logical limit: when the 5th device sees the stack hello messages originated by the stack members it reacts to this by disabling its own stack ports isolating itself from the stack.
The forwarding of frames between stack members likely use some form of encapsulation with a field 2 bits wide that points to the stack member. This is a guess on the stack implementation.
Hope to help
Giuseppe
03-01-2013 04:00 PM
When someone just regurgitates some documentation to answer a question I find it insulting, because I've already read the documentation and now have questions that require further elaboration.
Thanks for posting your findings and your insult.
I didn't bother running any "tests" on finding any limitation because with a 10 Gbps backplane, I didn't bother. I also know that the number of switches-per-stack limitation is hardware limited in the OS because of the stacking capabilities of the 3750 and 3850 switches.
So far, you are the first person who question the number of switches in a stack and which is why you didn't get any responses "in the wild". Cisco's Switching BU's respond to the question of stacking more than 4 switches in stack will always be "get a 3750".
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide