I have configured a PTP wireless network that requires RSTP protocol in order to manage redundance loops. Most of the loops are 4 points big, but there is a big one of more than 15 points. The most of the points are connected by PTP wireless radiomodems.
All switches are Layer 2 with RSTP protocol, but four of them with layer 3 that connect with root switch and root switch (green and yellow switches). Path cost are configured by defauld with 200000 in all STP enabled ports, but different configurations indicated in the picture, in order to balance the traffic in normal conditions (all points connected).
When we startup the network it shows convergence problems, that taked it down frequently, and we require to open some ports (or taking down some radiolinks) to take it up again.
After researching I found that I should change RSTP times at root switch according to the network size. After many tries I finally set Hello Time to 2 sec, Max Age to 40 sec, Forward Delay to 30 sec and Aging Time to 300 sec. This certainly make more stable the network, but it still goes down after some days of normal working.
Keep searching, I have found information about port flapping problem that appear to be a issue in RSTP working network when running over a wireless network. This happens when some radiolinks decrease its performance and goes down and up frequently. I think this probably is happening with the network as I have detected port flapping at root switch, also because I detected changes of the radiolink conditions over the time and convergence issues at the whole network. Also I have detected that some ports have been set "blocking" even when there is a opening at other point of the loop.
Other issue I have observed happens when the network is working correctly after converged with RSTP parameters, and in some moment, some of the switches and radios stop answering at ping request (using a network monitoring system), even when the equipment connected to them answer (PLC, radio, UPS, etc). This issue usually disappear after a while but it is seen as "no available" by the NMS and decrease the availability of the whole system.
At this point I wonder if there exists other considerations or checkpoint that I can review in order to improve the stability of the network.
For your reference and further information, I´m attaching the link with details of the complete network including IP configuration for all ethernet equipment, such as switches, radios, PLCs, HMI and UPS. Also there you can find the complete list of network devices indicating models and manufacturer. A second list shows the summary for switches configuration (ports and RSTP). Also I have attached the detailed configuration of the root switch (CISCO).
Please ask me if you require any further information.
Thank you in advance for your help, best regards,
Luis A. Cristóbal V.
About the solutions suggested and because of the network needs, I have sketched VLANs in the topology as showed partially in the picture attached. I have configured access ports (VPC) and trunk ports in between all switches. In this way, there are three trunking ports in CS_Callao switch, which allows all the VLANs.
Now my problem is that I need to use port E0/0 of CS_Callao switch to connect to any of the VLANs. In this stage I have tested configure E0/0 as access but it allow only one VLAN. If I configure it to trunk, I can´t use a VPC directly, it needs a trunk compatible device.
CS_Callao switch is a Cisco 3850, so it support IP routing, but all the VLANs need to be in the same subnet (10.25.44.XX), so I cannot configure different subnets for each VLAN and 3850 cannot route networks in the same subnet. At least I do not know how to do it.
I also review Private VLAN feature, but the issue is that each port connected should refer to a unique VLAN and this is not the case as you can see.
Do any of you, expert friends has an additional suggestion?
Thank you in advance!
Luis A. Cristóbal V.