1920 switches MSTP

[Dunky]

I am new to these switches, having previoulsy used Cisco and ProCurves.

I have three issues that are really causing me stress at present as I just cannot get them to work.  Rather than posting all three issues I will keep each post to a single issue.

There are seven 1920-24D-PoE+ switches in the configuration.

MSTP

There are a number of VLANs., and the edge switches have 2 fibres back to the main comms room switches (see attached).

I want to get VLANs 2-499 over one fibre and 500-599 over the other fibre.

I configured 2 instances and associated each with a range of VLANs as above.  Then in order to route the VLAN traffic over a specific fibre I amended the cost for the instance on the fibre ports on the edges switches which to my mind should have done what I was trying to acheive.

However, when I mirrored the fibre interface and looked on wireshark, all traffic was going over one fibre regardless of the source VLAN. 

The relevant part of the config is below (from switch-04).

#
#
stp region-configuration
region-name MSTREGION-AAA
instance 1 vlan 2 to 499
instance 2 vlan 500 to 599
active region-configuration
#
interface GigabitEthernet1/0/25
description Trunk to -00/g25
port link-type trunk
port trunk permit vlan all
flow-control
loopback-detection enable
loopback-detection control enable
loopback-detection per-vlan enable
stp instance 1 cost 5
mirroring-group 1 mirroring-port both
#
interface GigabitEthernet1/0/26
description Trunk to -02/g25
port link-type trunk
port trunk permit vlan all
flow-control
loopback-detection enable
loopback-detection control enable
loopback-detection per-vlan enable
stp instance 2 cost 5
#

Thanks
Steve

[Richard Brodie_1]

It's hard to say without all the configurations, but you have one hop to the switch to the root from switch 4 on g25, and three via g26. Are you sure you have the total post cost lower by changing just the costs on switch 4? It might be easier to increase the cost of the route you want to block.

[Dunky]

Very good thinking there RIchard!

If I go to Network>MSTP-Port Sumary and click on the fibre interfaces (25 and 26), the cost shows as 5 against the instance where I have set it to 5 and 20 on the other....

On switch -04, this is what it shows:

---[Port25(GigabitEthernet1/0/25)][FORWARDING]----
Port Protocol :enabled
Port Role :CIST Root Port
Port Priority :128
Port Cost(Legacy) :Config=auto / Active=20
Desg. Bridge/Port :0.2c23-3a7f-d506 / 128.25
Port Edged :Config=disabled / Active=disabled
Point-to-point :Config=auto / Active=true
Transmit Limit :10 packets/hello-time
Protection Type :None
MST BPDU Format :Config=auto / Active=legacy
Port Config-
Digest-Snooping :disabled
Num of Vlans Mapped :1
PortTimes :Hello 2s MaxAge 20s FwDly 15s MsgAge 0s RemHop 20
BPDU Sent :6
TCN: 0, Config: 0, RST: 0, MST: 6
BPDU Received :4405
TCN: 0, Config: 0, RST: 0, MST: 4405

Instance Cost Priority
1 5 128
2 20 128

----[Port26(GigabitEthernet1/0/26)][DISCARDING]----
Port Protocol :enabled
Port Role :CIST Alternate Port
Port Priority :128
Port Cost(Legacy) :Config=auto / Active=20
Desg. Bridge/Port :32768.2c23-3a81-47c6 / 128.25
Port Edged :Config=disabled / Active=disabled
Point-to-point :Config=auto / Active=true
Transmit Limit :10 packets/hello-time
Protection Type :None
MST BPDU Format :Config=auto / Active=legacy
Port Config-
Digest-Snooping :disabled
Num of Vlans Mapped :1
PortTimes :Hello 2s MaxAge 20s FwDly 15s MsgAge 0s RemHop 18
BPDU Sent :1061
TCN: 0, Config: 0, RST: 0, MST: 1061
BPDU Received :7043
TCN: 0, Config: 0, RST: 0, MST: 7043

Instance Cost Priority
1 20 128
2 5 128

Is the problem that g26 is DISCARDING - this seems to apply to all instances and I cant get my head around why it isnt shown per instance.

[16again]

I'd start off by making  switch00 primary root for instance 1 and secondary root for instance 2.
And sw01 primary root for instance 2 and secondary root for instance 1.
This way, you already get most of your required load sharing without messing with interface STP priorities

[Dunky]

I initally tried configuring switch-00 as primary for both instances and switch-01 secondary for both instances.  However it didnt appear to work as switch-03 was the root for both as it had the lower MAC address!  I have now made that switch -00 and assumed there was a bug as to why setting the instance to primary and secondary did not work.

Do you know why g26 is showing as DISCARDING and why it doenst appear to be possible to see what instance is DISCARDING - I think this may be part of the problem?

[16again]

One word struck me when browsing your log output.  CIST !
This suggests, MST isn't working as a single region, which might explain why  your traffic isn;t split as you intended

[Dunky]

Could you elaborate please.

I created an MST Region on every switch with the same name and revision number 0.

Syslog indicates that MSTP is running but I just cannot see any way that you can see the status of MSTP per instance ( such as yuou can with PVSTP on Cisco).

All that shows is the detailed info for the ports that I posted previously which is not per instance.

[16again]

I figured CIST meant MSTP wasn't working properly, since it's also used to connect non MSTP switches
But you're only seeing instance 0 in web GUI, so maybe it's OK afterall.
To really troubleshoot, enable telnet, and use _cmdline-mode-on  to get full shell.   (unsupported, might void warranty / support)  
For CLI manuals , see HP and h3c.com website

[Dunky]

Cheers, I will have a look at what debugging I can do via the command line when I am next in the office on Monday.

Still find it unbeleivable that I cant see the port status for each MST instance.

[Dunky · Accepted solution]

Well,finally got it working.

Reset everything back to defaults

Created MSTP region with Revision 1 this time instead of 0

On each of the three edge switches, set the cost on one of the fibre ports on instance 1 to 500 and instance 2 to path cost to 500 on the other fibre port as was recommended in the thread.

Tested with wireshark and behaving exactly as expected except for broadcasts which alwasys seemed to go over the first fibre, but I am not too worried about that.

[16again]

The core switch should sent out broadcasts to both fibers.  On the access switch, one receiving port is in STP-block mode, this is where packets are discarded

[Dunky]

Sorry, I meant broadcasts from a client patched to one of the edge switches.

[16again]

The edge switch shouldn't send out broadcast on its blocked port.  However, the broadcast sent on its forwarding fiber is copied by the core switch to all it's interfaces , so it also ends up on the fiber with blocked port.