4

I have cisco nexus switches in vPC and they are connected to HP c7000 blade center switches and all my c7000 blade server has Linux Bridge is running, i believe something went wrong somewhere in Linux Bridge configuration which created loop and it block trunk ports, now question is how do i find out where that loop happened and how do i recover from loopguard state?

I have multiple VLAN on switch but only VLAN 27 get stuck in loop, I have tried debug spanning-tree but i didn't see anything in debug.

If i shutdown servers do you think that will stop loop?

swt-tor2(config)# show spanning-tree vlan 27

VLAN0027
  Spanning tree enabled protocol rstp
  Root ID    Priority    8219
             Address     0023.04ee.be01
             Cost        3
             Port        5094 (port-channel999)
             Hello Time  2  sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    32795  (priority 32768 sys-id-ext 27)
             Address     00c8.8bbd.557c
             Hello Time  2  sec  Max Age 20 sec  Forward Delay 15 sec

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- --------------------------------
Po3              Root FWD 1         128.4098 (vPC) P2p
Po11             Desg FWD 1         128.4106 (vPC) P2p
Po12             Desg FWD 1         128.4107 (vPC) P2p
Po13             Desg BKN*1         128.4108 (vPC) P2p *LOOP_Inc
Po14             Desg BKN*1         128.4109 (vPC) P2p *LOOP_Inc
Po15             Desg FWD 1         128.4110 (vPC) P2p
Po16             Desg FWD 1         128.4111 (vPC) P2p
Po21             Desg FWD 1         128.4116 (vPC) P2p
Po22             Desg FWD 1         128.4117 (vPC) P2p
Po23             Desg BKN*1         128.4118 (vPC) P2p *LOOP_Inc
Po24             Desg BKN*1         128.4119 (vPC) P2p *LOOP_Inc
Po25             Desg FWD 1         128.4120 (vPC) P2p
Po26             Desg FWD 1         128.4121 (vPC) P2p
Po31             Desg FWD 1         128.4126 (vPC) P2p
Po32             Desg FWD 1         128.4127 (vPC) P2p
Po33             Desg BKN*1         128.4128 (vPC) P2p *LOOP_Inc
Po34             Desg BKN*1         128.4129 (vPC) P2p *LOOP_Inc
Po999            Root FWD 1         128.5094 (vPC peer-link) Network P2p

Update - 1

enter image description here

We have very basic switches in bladecenter nothing fancy.

on server side i have configured bond0.27 and on top of bond0.27 i have configured bridge interface (which has STP off)

# Openstack Octivia Managment Interface
DEVICE=br-lbaas
NAME=br-lbaas
BOOTPROTO=static
ONPARENT=yes
TYPE=Bridge
ONBOOT=yes
NM_CONTROLLED=no
DELAY=0
STP=no
IPADDR=172.27.8.21
NETMASK=255.255.248.0
ETHTOOL_OPTS="-K ${DEVICE} gso off sg off tso off tx off"

Directly connected bladecenter switch ports are loop-BLK* currently and we have many VLAN only VLAN 27 specific in loop-BLK*, other VLAN traffic flowing normal

This is port configuration connected to blade switches ( Port-Channel13 is currently in loop protection mode which you can see above) 

interface port-channel13
  description "connected to blade" 
  switchport mode trunk
  switchport trunk allowed vlan 10-12,20-22,27-32,40,50,100,200
  speed 10000
  vpc 13

I believe when i configured some openstack neutron networking that time something happened, i don't know where that BPDU comes, I don't think Linux server sending any BPDU in this scenario.

Do you think i should remove VLAN 27 from all switches and reconfigure?

Update -2

Here is the output of spanning-tree https://pastebin.com/F7mGdzEX

Here is the log output which you requested.

swt-tor1-010101-1-3-hd# show logg | include LOOP
2019 Jun 28 10:34:30 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel13 on VLAN0027.
2019 Jun 28 10:44:16 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel23 on VLAN0027.
2019 Jun 28 10:48:13 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel33 on VLAN0027.
2019 Jun 30 22:23:29 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel34 on VLAN0027.
2019 Jul  2 22:25:34 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel24 on VLAN0027.
2019 Jul  2 22:26:36 swt-tor1-010101-1-3-hd %STP-2-LOOPGUARD_BLOCK: Loop guard blocking port port-channel14 on VLAN0027.

Here is the HP 6120XG switch spanning-tree output: https://pastebin.com/LE05x68P

Improve this question
12
  • How does that Linux Bridge on the Blade Chassis behave? Is it a bridge talking STP (... at what priority?) ? Does it drop BPDUs? Does it bluntly forward BPDUs? Is the bridge using the LACP interface (a.k.a. "bond X"), or is it binding to the LACP member ports? What does show spanning-tree vlan 27 detail reveal? Please show the configuration of both port-channel and its members, and the other spanning-tree configuration bits. – Marc 'netztier' Luethi Jul 3 '19 at 22:17
  • ... and then, by all means, also investigate what those connectivity modules ("Virtual Connect" or one of the switch modules) in the blade chassis actually do in the spanning-tree game and what they do with BPDUs. – Marc 'netztier' Luethi Jul 3 '19 at 22:25
  • What is intriguing is that a port being put in Loop_Inc state must've been receiving BPDU's before; in extenso: it must've been a non-designated port, but a root or alternate port. It is being put into Loop_Inc state until it starts receiving BPDUs again. (community.cisco.com/t5/routing/…). If this is supposed to be an edge port connecting to a "server" system (well, Virtual Connect or Switch Module aren't exactly "servers"), then why was it receiving BPDUs before? – Marc 'netztier' Luethi Jul 3 '19 at 22:38
  • @Marc'netztier'Luethi I have updated my question, hope it help you to understand – Satish Jul 3 '19 at 23:46
  • 2
    That HP switch - if at all - most certainly runs MST, not (rapid)PVST. With MST, it can't have any notion of BPDUs for one VLAN, but not another; all BPDUs are sent on the given trunk's default VLAN as untagged frames. If all forms of STP are disabled on the HP Switch, you must find out what it does with BPDUs - pass them on (good) on or drop them ( bad!), and if it does so reliably. – Marc 'netztier' Luethi Jul 4 '19 at 5:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.