Switching

I am a little confused about the spanning tree state on an EX4200 VC, running 11.4R5.5.

{master:0}
cjc@dmz4200> show spanning-tree bridge       

STP bridge parameters
Context ID                          : 0
Enabled protocol                    : RSTP
  Root ID                           : 32768.88:e0:f3:77:53:81
  Hello time                        : 2 seconds
  Maximum age                       : 20 seconds
  Forward delay                     : 15 seconds
  Message age                       : 0
  Number of topology changes        : 20
  Time since last topology change   : 3661899 seconds
  Topology change initiator         : xe-0/1/0.0
  Topology change last recvd. from  : 88:e0:f3:74:51:6a
  Local parameters
    Bridge ID                       : 32768.88:e0:f3:77:53:81
    Extended system ID              : 0
    Internal instance ID            : 0

{master:0}
cjc@dmz4200> show spanning-tree interface 

Spanning tree interface parameters for instance 0

Interface    Port ID    Designated      Designated         Port    State  Role
                         port ID        bridge ID          Cost
ge-0/0/0.0     128:513      128:513  32768.88e0f3775381     20000  FWD    DESG 
ge-0/0/47.0    128:560      128:560  32768.88e0f3775381     20000  FWD    DESG 
xe-0/1/0.0     128:609      128:609  32768.88e0f3775381      2000  FWD    ROOT 
xe-0/1/2.0     128:611      128:611  32768.88e0f3775381      2000  FWD    DESG 
ge-1/0/0.0     128:625      128:625  32768.88e0f3775381     20000  FWD    DESG 
ge-1/0/47.0    128:672      128:672  32768.88e0f3775381     20000  FWD    DESG 
xe-1/1/0.0     128:721      128:776  32768.50c58dac4c81      2000  BLK    ALT  
xe-1/1/2.0     128:723      128:723  32768.88e0f3775381      2000  FWD    DESG 

So what I'm confused about is that the "show spanning-tree bridge" output says that this switch is the root bridge, yet the per-interface output indicates that there are a ROOT and ALT ports. Also, the bridge ID for the ROOT port is the switch itself? Whereas the ALT port is what I would expect, except that it seems to contradict the idea that this switch is the root bridge.

I think my RSTP on this switch is in some messed up state? When I turn on traceoptions for rstp, I see sucpicious stuff like,

Oct  2 11:26:59.532378 PISM: Port xe-1/1/0.0: IMPOSSIBLE EVENT/STATE Combination Occured
Oct  2 11:26:59.532396 PISM: Event routine returned FAILURE
Oct  2 11:26:59.532414 MSG: RstPortInfoMachine function returned FAILURE!
 
Oct  2 11:26:59.532441 RECV: PortReceiveStateMachine Action returned FAILURE!
                                        
Oct  2 11:26:59.532467 MSG: RstPortReceiveMachine function returned FAILURE!
                                        
Oct  2 11:26:59.532493 MSG: RstHandleInBpdu function FAILED!
                                        

have you tried rebooting the switch?

That seems a bit drastic.

In the end, I just stopped RSTP and restarted it.

1) I manually disabled the alternate root link so there would be no loop.

2) I turned off RSTP,

  # delete protocol rstp

  # commit

3) I re-enabled it,

  # set protocol rstp

4) Re-enabled the link disabled in (1).

And things look much better now:

cjc@dmz4200> show spanning-tree interface    

Spanning tree interface parameters for instance 0

Interface    Port ID    Designated      Designated         Port    State  Role
                         port ID        bridge ID          Cost
ge-0/0/0.0     128:513      128:513  32768.88e0f3775381     20000  FWD    DESG 
ge-0/0/47.0    128:560      128:560  32768.88e0f3775381     20000  FWD    DESG 
xe-0/1/0.0     128:609      128:610   4096.50c58dac4c81      2000  FWD    ROOT 
xe-0/1/2.0     128:611      128:611  32768.88e0f3775381      2000  FWD    DESG 
ge-1/0/0.0     128:625      128:625  32768.88e0f3775381     20000  FWD    DESG 
ge-1/0/47.0    128:672      128:672  32768.88e0f3775381     20000  FWD    DESG 
xe-1/1/0.0     128:721      128:776   4096.50c58dac4c81      2000  BLK    ALT  
xe-1/1/2.0     128:723      128:723  32768.88e0f3775381      2000  FWD    DESG

{master:0}
cjc@dmz4200> show spanning-tree bridge

STP bridge parameters
Context ID                          : 0
Enabled protocol                    : RSTP
  Root ID                           : 4096.50:c5:8d:ac:4c:81
  Root cost                         : 2000
  Root port                         : xe-0/1/0.0
  Hello time                        : 2 seconds
  Maximum age                       : 20 seconds
  Forward delay                     : 15 seconds
  Message age                       : 1
  Number of topology changes        : 1
  Time since last topology change   : 257832 seconds
  Topology change initiator         : xe-0/1/0.0
  Topology change last recvd. from  : 88:e0:f3:74:51:6a
  Local parameters
    Bridge ID                       : 32768.88:e0:f3:77:53:81
    Extended system ID              : 0
    Internal instance ID            : 0

Make sure you are running recent code, like 11.4 / 12.1 or newer.  There were several issues with STP not dealing with minor changes properly (e.g creating/deleting vlans) in older code and required some additional steps to get it to work properly.

What you can try in the future:

1) commit full

- This is usually enough to kick STP into behaving properly.  Basically, it forces the daemons to completely reprocess their configuration instead of just signalling a subset of daemons to pick up some changes.  Not disruptive.  Can even do it without a change. 

2) restart ethernet-switching 

- You could try using the "soft" option, but I doubt it will work since that is comparable to what a normal commit does.  If it doesn't, I would suggest using the "immediately" option to get it restarted as quickly as possible.  The "gracefully" option tries to be nice and start a new process before killing the old one but I have had it incur a longer disruption than "immediately" does.