@AmmarAlbis wrote:
For the first question, It seems I might have misunderstood something. I thought that when the root port link fails, it opens the alternate port and uses it as a new path. But, after reading the protocol again, the alternate port is just blocked. It doesn't just open when the root port fails. That's why we need to recalculate the STP. It never had an alternate path in mind in the first place. Is this conclusion right?
Hopefully, my prior reply will demonstrate that an alternate role pole doesn't necessarily guarantee it will become a root port, if the root port fails.
BTW, have some fun, experiment with the PT attachment, for example, look at what happens when I add another link to between some of the switches:
@AmmarAlbis wrote:
Second question, so basically their are timers and hop counts. So if a switch learns of a new better root bridge, it resets the timer. If this root bridge remain the same for a certain time, then this root bridge is the best(lowest BID) and all switches also chose it within this time. As for Rapid-STP it has a counter implementation to reduce the time for convergence. Is this conclusion correct?
Again, if the root bridge fails, that's a big deal, especially with non-rapid STP. I believe that, per standard, takes 50 seconds.
Off the top of my head, I do not try to remember all the things that rapid-STP does differently, to work "rapidly". Also, in the Cisco world, besides Cisco extending their STP implementations to be per VLAN (PVST) they provided multiple "enhancements", some, but not all, incorporated into the rapid-STP standard. Also, I believe, (especially with Cisco?), STP and rapid-STP "terms", might be the same words, but deep down, they don't work exactly the same. Unless going for a certification test, or troubleshooting STP, the two things to remember, try to avoid complex L2, if L3 can be used instead, and if using STP, if possible, use a rapid variant (which includes MST). Also, learn about STP feature enhancements, as some of they are very useful.
In other words, things like insuring certain port roles, or changing SPT parameter to improve "speed" of convergence, aren't usually as important as other higher level design considerations. For example, primary and secondary root selection, or, perhaps, using Etherchannel, for multiple L2 links between switches, than STP for redundancy, or limiting scope of broadcast and unicast flooding, i.e. pruning, etc.
As you mention being a newbie to networking, often, conceptionally, networking isn't too difficult, but to do it well often requires a very broad knowledge of networking.
I've attached a PT file.
When you load it, and STP converges, it should look like:
BTW, switches are running PVST (non rapid), and top switch is configured as root.
The orange interfaces are blocked.
If I shut the top left active link, when STP again converges, you get:
Hmm, the alternate link between the top and left most switched not used, port on bottom switch, to right switch unblocked, why?
If we look at before and after port roles for the left and bottom switches, we see:
Left switch:
Before After
Interface Role Sts Interface Role Sts
---------------- ---- --- ---------------- ---- ---
Gi1/0/3 Altn BLK Gi1/0/3 Altn BLK
Gi1/0/1 Root FWD
Gi1/0/2 Desg FWD Gi1/0/2 Root FWD
Bottom switch:
Before After
Interface Role Sts Interface Role Sts
---------------- ---- --- ---------------- ---- ---
Gi1/0/1 Root FWD Gi1/0/1 Desg FWD
Gi1/0/2 Altn BLK Gi1/0/2 Root FWD On the left switch, when I shut the root port, the alternate port stayed the an alternate and the designated port became the new root, so some form of STP calculation was performed, correct? I.e. the alternative didn't automatically become the new root.
On the bottom switch, the root port became a designated port and the alternative port became the root port, but no links were changed, so some change to the topology, forced a STP calculation, correct?
BTW, if you watch the switches, you'll see ports going into listening/learning roles, during the transition, which is expected.
So, perhaps STP isn't just as simple of alternative role ports always becoming a root when the root port fails. Additionally, you can have multiple alternatives.
Oh, if you're wondering why the left most switch link to the top most switch didn't become the new root, that was because all links but it are running at gig, while it is running at 10 Mbps. I.e. higher STP link cost, which impacts selection of best path to root, when doing STP best path calculation.
If you're also wondering why you might actually have a topology as above, perhaps the two gig links are in the same street conduit (i.e. subject to a "backhoe" cut) while the 10 Mbps link is p2p WiFi running across the street between the two switch locations. (In the real-world, whether your redundant fiber links share a conduit, or whether your two independent service providers use the same POP, can matter for redundancy.)
