Resilience of Cat. 2960-s FlexStack Switch

Unanswered Question
May 3rd, 2012
User Badges:

Hello,

I wish your everything is going well all the time.

I have a question for Cisco Cat.2960-s FlexStack switches which are installing on our sties.

Two of 2960-s Stack swithces as access switch and two of Cisco ME 3600X Series as distribution layer switches are to be istalled in our sites.

In case of two stack switches, One is will be a Master and the other one will be a member logically, as you know.

So, if the master fails, the other one automatically becomes the stack master following a well-documented election process.

Now, it is my question.

How long takes to be a stack master from a member switch ?  I cant find it on white paper of Cat.2960-s flexstack .

And also, I heard that sometimes a member switches don't election process when the master fails as a result, all stack members become

a panic.  Is that really right ?

In addition, I heard that the stack switches have many troubleshooting points than stand alone switches.

I really wanna know if the stack switches are good solution for resilience of huge network site.


I'm waiting an answer from  those who have experience of maintenance or installation.

Thanks for reading my post.


Best Regards,


From Bob.


Message was edited by: Soo Hyun Song

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 2 (1 ratings)
Loading.
Sergey Fer Fri, 05/04/2012 - 01:22
User Badges:
  • Bronze, 100 points or more

Well, the FlexStack architecture is a bit tricky. It uses technology which is called HW Drop Table internally to avoid loops and in some cases it resembles STP That is why in some cases it will take significant time to recover. It may take 1-2 seconds for stack to converge. Stack master election algorithm is deterministic and it must work in a bugless environment.

Leo Laohoo Fri, 05/04/2012 - 18:36
User Badges:
  • Super Gold, 25000 points or more
  • Hall of Fame,

    The Hall of Fame designation is a lifetime achievement award based on significant overall achievements in the community. 

  • Cisco Designated VIP,

    2017 LAN, Wireless

Hey Bob,


Out of curiousity, what are you trying to do?  I hope you are NOT installing 2960S in a data centre.

shsong21 Tue, 05/08/2012 - 21:08
User Badges:

Hello leolaohoo,

Thanks for your comment.

Do you have experience of maintenance or running ?

Im trying to install them as an access switch in my customer sites and the switches are to be sent video , data,

alarm signals to backbone switches.

This system is intergrated security system such as CCTV, IDS, ACS.

If you know something, plz let me know.


Regards,

Leo Laohoo Tue, 05/08/2012 - 21:39
User Badges:
  • Super Gold, 25000 points or more
  • Hall of Fame,

    The Hall of Fame designation is a lifetime achievement award based on significant overall achievements in the community. 

  • Cisco Designated VIP,

    2017 LAN, Wireless

How many of these devices go to the switch (stack)?

shsong21 Wed, 05/09/2012 - 02:12
User Badges:

It's different as per sites so I cant tell in detail here.... By the way, why u need such that information ?

Im asking what is the problem of stack switches.

I heard that if the master fails, sometimes other members(slave) dont become a master.

In the result, all stack switches are not work. Is that really right ?

I want some bugs of stack switches.

Do u know that ?


Regards,

Leo Laohoo Wed, 05/09/2012 - 15:51
User Badges:
  • Super Gold, 25000 points or more
  • Hall of Fame,

    The Hall of Fame designation is a lifetime achievement award based on significant overall achievements in the community. 

  • Cisco Designated VIP,

    2017 LAN, Wireless

Im asking what is the problem of stack switches.

I heard that if the master fails, sometimes other members(slave) dont become a master.

Never heard or seen this before.  I have 2960S and 3750/3750E/3750X in my network and I've never seen this happen before.  Everytime the master fails, the next member with the highest switch priority OR the lowest MAC address will become the switch master.  If you want to stack them then ensure that each stack member has a valid uplink by configuring EtherChannel.


By the way, why u need such that information ?

If you are concerned about stacking and the recovery, then consider this:


Presume you have a stack of four 2960S-48LPD (maximum allowed to be stack by this model) and you have 40 CCTV.  (I'm going to presume that each CCTV camera has ONE uplink.) If you want redundancy, you do not want to put all of the cameras to one switch member, instead you want to distribute them evenly across the stack.  Plus you also want to make sure that, say, a sector of eight goes to the same switch either.  You want to spread across the different stack member.


How about power?  Because the 2960S does not have redundant power supply (it does, however support DC power for redundancy) so you either want to plug two 2960S to mains and two to UPS or all four to UPS.  The UPS have to be tested each year.


This is where the 3750X and 2960S sets the difference.  The 3750X supports power stacking which would be great if a power supply of a switch member fails and one will provide extra power (even with PoE+ is enabled).

shsong21 Wed, 05/09/2012 - 18:00
User Badges:

Hello,

Thanks for your comment.

By the way, I heard and seen this in my customer network so I have requested 'case open' to Cisco Korea

but they have not any response to us. and also, the FlexStack cable raise a problem very often.

How about this ? never heard and seen before ?

You know what, the biggest problem of stack switch is that if the master fails, all members fail at the same time.

What do u think of this ?

One more, 2960-s switches connected with RPS(Redundant Power System) for resilience.

Leo Laohoo Wed, 05/09/2012 - 18:16
User Badges:
  • Super Gold, 25000 points or more
  • Hall of Fame,

    The Hall of Fame designation is a lifetime achievement award based on significant overall achievements in the community. 

  • Cisco Designated VIP,

    2017 LAN, Wireless

You know what, the biggest problem of stack switch is that if the master fails, all members fail at the same time.

What do u think of this ?

Like I said in my previous post, I have never seen this before and I don't believe this is possible.  What do you mean "all members fail the same time"?


I mean when you said that all the members will not pass traffic upstream?  This is only possible if the uplinks were not configured correctly.


As to the possibility that if you have a stack of 2960S with the uplinks configured correctly and the master fails and the stack members sits and do nothing, I do not believe this is true.


I have 2960S stacks.  I have them configured as etherchannel.  I have the stack master fail.  And the rest of the members continue to pass traffic.  I did not know the stack master failed until we received a case that a number of clients can't log-on to the network. 

paolo bevilacqua Thu, 05/10/2012 - 00:48
User Badges:
  • Super Gold, 25000 points or more
  • Hall of Fame,

    Founding Member

Update to latest IOS, and try again.

Actions

This Discussion