Outage Cause By changes on STP

[JoeGo88]

I am experiencing  problem with STP on the below diagram , every time Changes happen at STP over the switches it cause network outage for the network behind the router as well routing table 

Sw 1 , 2 ,3 ,4 directly connected to each other with trunk port holding vlan x1 and x2 

Eigrp is configured between R 1,2,3,4  

SW1 configured as root bridge to vlan X1 

 there  private network behind each router 

when i lost sw2 , the connection between R3 and R4 drop for 15s which cause network outage even the direct connection between them still up , i don't why ?

can i prevent this behavior ?

Thanks 

[paul driver]

Hello
Does the diagram you've posted represent the actual network? 
Can you  post in an attached file the following outputs:

sh spanning-tree detail
sh interface trunk
sh vlan brief

[JoeGo88]

Thanks Paul 

see attached 

[MHM Cisco World]

this normal for this design, 
I think SW2 or SW1 is the root bridge and hence when SW2 is down 
the SW3 and SW4 start new root election and the link is BLK->FWD and this take time, the time what you mention that the connection is loss. 

solution, try make each SW have other link to Root SW.

[MHM Cisco World]

this my idea, and to make sure that this link not elect as root port, you can change it Cost to be higher than link interconnect SW's. 

now if SW2 is down the SW3,SW4 have other link to Root SW (SW1 or SW2). 

[JoeGo88]

if i change vlanx1 port priority per interface on the trunk port at sw3 facing sw4 while the root bridge is sw1 once sw2 down is sw3 and sw4 will start election and will face connection loss 

[MHM Cisco World]

friend if Root is SW1 or SW2 or SW3 or SW4 you will face same issue, the change of priority is not solution neither change root SW, 
what you need is add other link to change the network topology as I show above 

with this link always always there is path to root SW whatever it will be.
if SW2-SW3 is down 
SW3 and SW4 can take red path to SW1 and SW2 no new root elect 

if the SW1-SW2 is down the SW1 have red path to SW2 SW3 SW4 and hence no new root elect

if the SW4-SW3 down the SW4 have red path to SE2 SW3 SW1 and hence no new root elect

Note:- I prefer you change the cost/priority of red link to make it less prefer that link interconnect SW. 
red link is use only as failover link not as primary link. 

[JoeGo88]

Thank you 

[MHM Cisco World]

You are so welcome 

[balaji.bandi]

Daisy chaining the switches you see this known problem. as others asked is this LAB or real setup.

If this is real, what is needed to connect the daisy chain? are they too far ? (if they are close enough) - you need to change the network design by understanding the failure cases.

coming back to your problem :

making SW1 as root bridge you see this issue.

So you need to plan where the STP root bridge needs to be and alternative which can act as a root bridge if the root failed, as alternative root. by setting priority

in your case (if you not able to make changes in the physical topology, I would suggest to use SW2 or SW3 as root bridge with priority) - this will minimise the impact of network.

in the trunk you have VLAN x1 and x2 passed to all switches, and you have only VLAN x1 allowed as an access switch - I take this as an example, and you may have more VLAN is this correct?

[JoeGo88]

This sample of live production network , there is more than 4 switches and 4 routers connected , 

you may have more VLAN is this correct? yes correct 

i have live camera feed behind each router so when STP changes happen i lost this feed for 20s which we consider and critical outage , i tried to make sw1as primary root and sw4 as secondary to avoid this outage but still the same issue as well , r1 and r4 on two different location each router have  full network behind including fw to provide failover over ipsec internet incase any failure happened on direct connection between them , i knew 20 sec is normal for this design but i am looking to decrease this time as much as i can with doing any physical changes on the topology 

thanks 

[balaji.bandi]

I have asked other question:

If this is real, what is needed to connect the daisy chain? are they too far ? (if they are close enough) - you need to change the network design by understanding the failure cases.

you need to post a complete network for us to suggest what we can here.

i would suggest having SW2 and SW3 as root bridge closest as per your topology.

if you think this is very critical you need to arrange an alternative path and redundant link part of business justification and also consider rapid stp for faster convergence. (depends on model of the switch and ios code running).

[James Hawkins]

From the diagram it would appear that there are no redundant links. If this is really the case then Spanning Tree is not required and could be disabled completed. 

That can be a little dangerous to do that as loops could be inadvertently introduced and cause network issues.

My other thought is can you not use Rapid Spanning Tree instead of classic STP. RSTP has been around for decades and offers very swift recovery from failures like the one you have described.