STP - Causing delays for new devices?

[Tom2]

Hello,

When connecting new devices to my network there seems to be a delay, possibly while STP goes from listening to forwarding?

I posted the question over in the Cisco forum as some of my kit is Cisco, the access layer however consists of HP 2650's https://supportforums.cisco.com/message/3308593#3308593

It was suggested there that I should set the access ports on my HP switches to "auto-edge-port" if they are not set already. I'm unable to find the settings and or confirm if they are already set as auto-edge, can anyone help?

Thanks for reading,

Tom

[Ardon]

Hi Tom,

You are looking for Admin Edge (if you are running RSTP 802.1w or MSTP 802.1s) which in effect will tell the switch you are not connecting another bridge/switch and making it safe for the port to go forwarding right away.
If, however you do happen to connect a Switch running STP to a port configured as Admin Edge, it will turn to a Operational Edge=no port.

Have a look at the OperEdgePort and AdminEdgePort below.

Status and Counters - CST Port(s) Detailed Information

Port : 1
Status : Down
BPDU Protection : No
BPDU Filtering : No
PVST Protection : No
PVST Filtering : No
Errant BPDU Count : 0
Root Guard : No
TCN Guard : No
MST Region Boundary : Yes
External Path Cost : 200000
External Root Path Cost : 0
Administrative Hello Time : Global
Operational Hello Time : 2
AdminEdgePort : Yes->configured
Auto Edge Port : Yes
OperEdgePort : Yes->Actual
AdminPointToPointMAC : True
OperPointToPointMAC : Yes
Aged BPDUs Count : 0
Loop-back BPDUs Count : 0
TC ACK Flag Transmitted : 0
TC ACK Flag Received : 0

In order to verify what is running both configured and effectively issue the: 'Show Span Detail' command


If you are running the legacy Spanning Tree (802.1d) they you should configure 'Span mode fast.

With Auto-Edge you will still have a 3 second delay as it will wait for 3 seconds to see if it receives a Spanning Tree BPDU.

[Tom2]

Hi Ardon,

We are running MSTP, how do I get the output that you placed in your post?


Output from Show Span Detail:

HP ProCurve G-B01(config)# sh sp de

Multiple Spanning Tree (MST) Information

STP Enabled : Yes
Force Version :

Switch Priority : 32768 Hello Time : 2
Max Age : 20 Forward Delay : 15

Topology Change Count : 3,004,932
Time Since Last Change : 3 mins

Root MAC Address : 001438-f89c80
Root Path Cost : 0
Root Port : 49
Root Priority : 0

HP ProCurve G-B01(config)#

Tom

[Tom2]

Hi Guys,

Can anyone help?

The delay is becoming a pain.

Tom

[Antonio Milanese]

Hi Tom,

there must indeed be a problem with (M)STP operations since you have abnormal TCs counters..please post result of following commands:

"sh ver"
"sh span all detail"
"sh span root-history ist"
"sh span instance ist"
"sh span debug-counters"
"sh span debug-counters ports all instance 0"
"show loggin -r stp"

maybe it's a classical CST vs MSTIs schizophrenia but you should check:

a) MSTP-operation compatibility i.e. firmwares baseline
b) correct MSTP config/instance/VLAN membership and definition
c) correct priorities/path cost calculation on CST (cisco change path cost on LACP channels HP do not!) especially on boundary ports
d) filter PVST(+) bpus on HP side but more generally i'm not inclined to run PVST on hp/cisco mixed LANs and use only MSTP
e) configure edge (postfast in cisco lingo) ports to avoid triggering costly and useless topology changes


Best regards,

Antonio

[Tom2]

Hi Antonio,

Thank you for your reply,

I have the output from the sh commands below although not all worked as expected, perhaps firmware differences?

All of the 2650 HP's are on the same firmware apart from a 5308xl which runs E.09.22

You mention configuring edge ports on the HP kit, how do you do this? I read previously that they should be automatically edge but

was unable to verify or specify..



HP ProCurve G-B01# sh ver
Image stamp: /sw/code/build/fish(f04)
Feb 8 2005 11:42:59
H.08.60
1219
Boot Image: Primary

HP ProCurve G-B01# sh span all detail

Multiple Spanning Tree (MST) Information

STP Enabled : Yes
Force Version :

Switch Priority : 32768 Hello Time : 2
Max Age : 20 Forward Delay : 15

Topology Change Count : 3,120,856
Time Since Last Change : 19 mins

Root MAC Address : 001438-f89c80
Root Path Cost : 0
Root Port : 49
Root Priority : 0

HP ProCurve G-B01# sh span root-history ist
Invalid input: root

HP ProCurve G-B01# sh span instance ist
IST Instance Information

Instance ID : 0
Mapped VLANs :

Switch Priority : 32768

Topology Change Count : 3,120,856
Time Since Last Change : 22 mins

Regional Root MAC Address : 001438-f89c80
Regional Root Priority : 0
Regional Root Path Cost : 20000
Regional Root Port : 49
Remaining Hops : 19
Designated
Port Type Cost Priority Role State Bridge
---- --------- --------- -------- ---------- ---------- -------------
1 10/100TX 200000 128 Designated Forwarding 001438-f8db80
2 10/100TX 200000 128 Designated Forwarding 001438-f8db80
3 10/100TX 200000 128 Designated Forwarding 001438-f8db80
4 10/100TX 200000 128 Designated Forwarding 001438-f8db80
5 10/100TX 200000 128 Designated Forwarding 001438-f8db80
6 10/100TX 200000 128 Designated Forwarding 001438-f8db80
7 10/100TX 200000 128 Designated Forwarding 001438-f8db80
8 10/100TX 200000 128 Designated Forwarding 001438-f8db80
9 10/100TX Auto 128 Disabled Disabled
10 10/100TX Auto 128 Disabled Disabled
11 10/100TX 200000 128 Designated Forwarding 001438-f8db80
12 10/100TX 200000 128 Designated Forwarding 001438-f8db80
13 10/100TX 200000 128 Designated Forwarding 001438-f8db80
14 10/100TX 200000 128 Designated Forwarding 001438-f8db80
15 10/100TX 200000 128 Designated Forwarding 001438-f8db80
16 10/100TX Auto 128 Disabled Disabled
17 10/100TX 200000 128 Designated Forwarding 001438-f8db80
18 10/100TX Auto 128 Disabled Disabled
19 10/100TX 200000 128 Designated Forwarding 001438-f8db80
20 10/100TX 200000 128 Designated Forwarding 001438-f8db80
21 10/100TX 200000 128 Designated Forwarding 001438-f8db80
22 10/100TX 200000 128 Designated Forwarding 001438-f8db80
23 10/100TX 200000 128 Designated Forwarding 001438-f8db80
24 10/100TX 200000 128 Designated Forwarding 001438-f8db80
25 10/100TX Auto 128 Disabled Disabled
26 10/100TX Auto 128 Disabled Disabled
27 10/100TX Auto 128 Disabled Disabled
28 10/100TX Auto 128 Disabled Disabled
29 10/100TX Auto 128 Disabled Disabled
30 10/100TX Auto 128 Disabled Disabled
31 10/100TX Auto 128 Disabled Disabled
32 10/100TX 200000 128 Designated Forwarding 001438-f8db80
33 10/100TX 200000 128 Designated Forwarding 001438-f8db80
34 10/100TX Auto 128 Disabled Disabled
35 10/100TX 200000 128 Designated Forwarding 001438-f8db80
36 10/100TX 200000 128 Designated Forwarding 001438-f8db80
37 10/100TX Auto 128 Disabled Disabled
38 10/100TX 200000 128 Designated Forwarding 001438-f8db80
39 10/100TX 200000 128 Designated Forwarding 001438-f8db80
40 10/100TX 200000 128 Designated Forwarding 001438-f8db80
41 10/100TX 200000 128 Designated Forwarding 001438-f8db80
42 10/100TX 200000 128 Designated Forwarding 001438-f8db80
43 10/100TX 200000 128 Designated Forwarding 001438-f8db80
44 10/100TX Auto 128 Disabled Disabled
45 10/100TX 2000000 128 Designated Forwarding 001438-f8db80
46 10/100TX 200000 128 Designated Forwarding 001438-f8db80
47 10/100TX Auto 128 Disabled Disabled
48 10/100TX Auto 128 Disabled Disabled
49 100/1000T 20000 128 Root Forwarding 001438-f89c80
50 1000SX 20000 128 Designated Forwarding 001438-f8db80

HP ProCurve G-B01# sh span debug-counters
Invalid input: debug

HP ProCurve G-B01# sh span debug-counters ports all instance 0
Invalid input: debug

HP ProCurve G-B01# show loggin -r stp
Invalid command: -r.


Tom

[Antonio Milanese]

Hi Tom,

firstly let me say that you are running very old versions and I strongly suggest you to update as soon as possible since there are numerous compatibility fixes/new features related to STP:

latest public 26xx image is H.10.83 and 53xx is E.11.21; make sure to read upgrade notes 'coz you'll be necessarily forced to go through some intermediate updates to the bootcode.

Secondly I suspect that you dont have a common MSTP config (or even running all switches indipendently) so you've a chatty RSTP boundaries with lot of external TCNs and recalculations..take a moment to read this must-read synthesis on the topic:

http://blog.ine.com/2010/02/22/understanding-mstp

for edge (host connecting) ports the cisco's "spantree portfast" equivalent on latest firmware is:
"spanning-tree n_port admin-edge-port"

When you have updated firmwares there are more troubleshooting commands (f.e. "sh span debug-counters")which can help you understand when and why TCs heppens!

Best regards,
Antonio

[Tom2]

Hi Antonio,

Ok, to be honest I have been waiting for an excuse to upgrade the firmwares. I think Easter will give me some opportunities to do just that.

I have also printed the blog post that you linked too, I haven't read it yet but it looks very informative so thanks for that.

I'll be back once I have the updated the switches.

Thanks,

Tom

[Gerhard Roets]

Hi Tom

1. Can you identify a port number and switch you see the issue on, and give us that.
2. Please post the output of "show logging"
3. If you can post your config minus keys and passwords it would be great.

Thanks in advance
Gerhard

[Tom2]

Hi Guys,

In advance of upgrading the firmware I think I might have stumbled across a potential issue/fix to at least some of my problems, Antonio you mention a 'common MSTP config' and this got me thinking. I read here:
http://cdn.procurve.com/training/Manuals/ProCurve-and-Cisco-STP-Interoperability.pdf

"Configure MST and enable MSTP globally. It is very important that all switch MST configurations match exactly. The name, revision, and instances VLAN mappings must be identical on all switches participating in MSTP configuration."

Looking at my config I have two switches [located in another building connected via direct fibre and on the same LAN but not connected directly to another switch with MSTP] that do not have the following lines in their line in their config that are common to all other switches participating in this MSTP config:
spanning-tree config-name "configname"
spanning-tree instance 1 vlan 1-37 39-41 202
spanning-tree instance 2 vlan 42-80

No loops exist with these switches, they are connected together via a single copper uplink with a single fibre back to the rest of our switched infrastructure. I think MSTP is in place to prevent accidental loops or for when/if we upgrade resiliency here.

I noticed that these switches are reporting the same root bridge switch MAC which is part of the MSTP loop in our main building. A Cisco switch separates the two with the following possibly redundant config:
spanning-tree mode pvst
spanning-tree extend system-id

Should I remove the MSTP config on these two outlying switches or give them a unique config-name? Would the unique name then cause them to elect a new root bridge from the two switches?

Common MSTP config:
spanning-tree
spanning-tree protocol-version MSTP
spanning-tree config-name "configname"
spanning-tree config-revision 1
spanning-tree instance 1 vlan 1-37 39-41 202
spanning-tree instance 2 vlan 42-80

I think I'm starting to get a better understanding!

Tom