Switch Stack Connections Go Down when Master Reboots

julito4589 · ‎08-11-2016

Hi,

I have to Cisco WS-C2960S-48TS-L on a stack. They're currently running IOS version is 15.0(2)SE6, C2960S-UNIVERSALK9-M.

The switch stack doesn't seem to be performing as expected.

Connections to devices behind the stack fail when member 1 is master and member 2 is slave.

Only when member 2 is master and member 1 is in ready /up state, does the stack work correctly and the devices behind the stack can be reached.

I've been looking at the documentation and at other posts in this community and the only command that I haven't issued in this stack is "switch n priority XX", which is recommended in some of the posts I have seen.

Can someone provide pointers on what might be missing on my configurations or what might be the cause of switch 1 never taking over as master?

Below is the output from a few of the stack related commands

#sh switch detail

Switch/Stack Mac Address : 1cde.a771.9280
Mac persistency wait time: Indefinite
                                           H/W   Current
Switch# Role   Mac Address     Priority Version State
----------------------------------------------------------
1       Member 1cde.a771.9280     1      1       Ready
*2       Master 1cde.a771.9600     1      1       Ready

         Stack Port Status             Neighbors
Switch# Port 1     Port 2           Port 1   Port 2
--------------------------------------------------------
1        Ok         Ok                2        2
2        Ok         Ok                1        1

#sh switch stack-ring speed

Stack Ring Speed : 10G
Stack Ring Configuration: Full
Stack Ring Protocol : FlexStack

#sh platform stack manager all
Switch/Stack Mac Address : 1cde.a771.9280
Mac persistency wait time: Indefinite
                                           H/W   Current
Switch# Role   Mac Address     Priority Version State
----------------------------------------------------------
1       Member 1cde.a771.9280     1      1       Ready
*2       Master 1cde.a771.9600     1      1       Ready

         Stack Port Status             Neighbors
Switch# Port 1     Port 2           Port 1   Port 2
--------------------------------------------------------
1        Ok         Ok                2        2
2        Ok         Ok                1        1

               Stack Discovery Protocol View
==============================================================

Switch   Active   Role    Current   Sequence   Dirty
Number                    State     Number     Bit
--------------------------------------------------------------------
1        TRUE    Member   Ready       124       FALSE
2        TRUE    Master   Ready       006       FALSE

                 Stack State Machine View
==============================================================

Switch   Master/   Mac Address          Version    Current
Number   Member                          (maj.min) State
-----------------------------------------------------------
1        Member    1cde.a771.9280          1.49        Ready
2        Master    1cde.a771.9600          1.49        Ready

Last Conflict Parameters

Switch Master/ Cfgd Default Image H/W # of    Mac Address
Number Member Prio Config   Type Prio Members
-----------------------------------------------------------------------

            Stack Discovery Protocol Counters
        Messages Sent                  Messages Recvd
        UP       DOWN                  UP        DOWN
--------------------------------------------------------
1: 0000068945 0000068945          0000068954 0000068954
*2: 0000069893 0000069893          0000069759 0000069759
3: 0000000000 0000000000          0000000000 0000000000
4: 0000000000 0000000000          0000000000 0000000000

             Misc Counters
   Counter                      Up          Down
---------------------------------------------------

Wrong Ver Number: Send:    0000000000    0000000000
Wrong Ver Number: Recv:    0000000000    0000000000
Missed Messages:           0000000000    0000000000
Orphaned Messages          0000000000    0000000000
Suppressed Messages        0000000000    0000000000
No Available Messages      0000000000    0000000000

Link Present               0000000000    0000000000
Link Not Present           0000000000    0000000000
Link RxReset               0000000000    0000000000
Link Sync Stuck Resets     0000000000    0000000000
Duplicates                 0000000001    0000000000
RAC Not OK Resets          0000000000
Switch# of last duplicate 0000000001
Sequence Number Failures   0000000000
RAC Not OK Resets          0000000000
Sync Not OK Resets         0000000000    0000000000
Switch# of last Failure:   256 Last Difference 0
Switch Number Conflicts    0
Stack Changes              7
Int Stack Link changes     0
Int Stack Link state       0x0
Reciprocal Efficiency Changes: Upgrade 0, Downgrade 0
             Resource Counters
-------------------------------------------
Chunk Alloc's        0000000010
Chunk Free's         0000000009
Enqueue Failures:    0000000000
Null Queue Failures: 0000000000
Chunk Alloc Errors: 0000000000

           Stack State Machine Counters
   Messages Sent              Messages Recvd
-------------------------------------------------
1: 0000000005                 0000000005
*2: 0000000000                 0000000000
3: 0000000000                 0000000000
4: 0000000000                 0000000000

julito4589 · ‎08-11-2016

Just to add some more info here. I currently don't have persistent MAC address enabled. So it's possible that when the master switch goes down, the slave switch takes over ok, but the arp table in the L3 switch that this stack uplinks to still has the MAC address of master switch. What are other folks doing for MAC address persistent settings?

Reza Sharifi · ‎08-11-2016

Hi,

It is recommend to use priority for the master switch (1-15) with 15 being the highest priority. This way, you know that a specific switch should always be the master. That said, the stack master should never lose its role unless the stack master is rebooted, powered off or fails, but to troubleshoot you may want to start with priority. Also, the default value for the stack master persistent is 4 minutes (I think). So, it maybe helpful to reduce it to a lower value (1 minute) for faster fail over.

HTH

julito4589 · ‎08-12-2016

Since it's Friday, I won't try any changes until early next week.

The current value for mac wait time persistency is set to "Indefinite"

#sh switch
Switch/Stack Mac Address : 1cde.a771.9280
Mac persistency wait time: Indefinite
                                           H/W   Current
Switch# Role   Mac Address     Priority Version State
----------------------------------------------------------
1       Member 1cde.a771.xxxx     1      1       Ready
*2       Master 1cde.a771.yyyy     1      1       Ready

I'm actually a little confused by this setting now and by the consequences it can create in other devices that this stack connects to. Based on the documentation "When a stack master is removed from the stack and a new stack master takes over, the default is for the MAC address of the new stack master to immediately become the new stack MAC router address."

So leaving this setting alone would imply that when the master goes down, the stack assumes the MAC address of the new master immeditely.

Is this bound to create ARP issues with, say, the switch that the stack uplinks to?

This is what I'm assuming is happening right now when my master changes as none of the devices behind the stack can be reached.

In your response though you recommend setting a specific value for the MAC address persistency for faster fail over. So I guess the question is what the documentation means by "immediately" when referring to the master's MAC address switch.

The more important question is whether I should change this setting at all or not. In the previous response, that was the recommendation, but it seems to contrast with what the documentation says.