Slow Printing/Scanning After HSRP Active Core switch failure

[Sihanu N]

Hi Experts,

We had two Core switches(HSRP active/standby mode)  in our environment. Recently Active Core switch failed due to some hardware problem and the now the HSRP active Core 2 switch with same running configuration. But after that the users in some vlans experiencing slow print/scan problem (those who try to print between two vlan), at the same time intra vlan printing/scanning works fine. Also the Root Bridge for the spanning tree(RPVST) is one Access Switch in the environment (which was previously like that).

1) Is there any spanning-tree related issue affecting in this case (any thing loop)? If then, how we can check this?

2) Is there any thing related to configuration mismatch with previous core 1 switch?

Any advice and suggessions regarding the issue are appreciable.

Regards,

Sihanu N

[glen.grant]

  The spanning tree root should always be on your core switches , primary root would be hsrp active side and secondary root would be the hsrp standby switch.  If there is a built in loop where one interface is getting blocked its possible you have some weird traffic patterns coming out of the access switch. Check for any interface errors on your uplinks.  I would try to fix the spanning tree root issue during a maintenance window.

[Sihanu N]

Hi Glen,

Many thanks for the reply. From the core switch the cost for reaching the root bridge is 3008 for all vlans.

1) Please suggest whether this value is a permissible one?

2) Is there any troubleshooting steps for finding the reason for the intervlan printing issue?

Also we are going to change during the next maintenance schedule. But we need to resolve it temporarily upto that maintenance shedule.

Regards,

Sihanu N

[Vivek Ganapathi]

Hello Sihanu,

As Glen suggested, the first option should be to fix your root bridge. Below is a strategy which i follow

1) Run HSRP on CORE - CORE1 being active, CORE 2 being standby.

2) Even VLANs on CORE 1 with Root bridge priority lower (means preferred for even vlans).

3) Odd VLANs on CORE 2 switch Root bridge priority lower (means preferred for odd vlans).

You must hardcode the spanning-tree parameters or else, it would prefer a switch with the least MAC address in your network. Looks to me like, when you primary CORE went down, it opted the one with the least MAC address as your root bridge.

Also, looks like you have a spanning-tree uplinkfast somewhere as i notice your cost to be 3000+ which generally happens when you run in uplinkfast. Well, this is automatically set as soon as you enable uplinkfast. But a proper planning needs to be exercised when you enable uplinkfast or else this feature wouldn't much benefit you.

To fix the issue, you need to configure your STP root bridge election parameters manually. Below is an example

spanning-tree vlan 2, 4, 6, 8 priority 4096 on core 1

spanning-tree vlan 3, 5,7,13 priority 8192 on core 1

spanning-tree vlan 3, 5,7,13 priority 4096 on core 2

spanning-tree vlan 2, 4, 6, 8 priority 8192 on core 2

Also, you need to adjust your HSRP parameters along with it. You should have VLAN 2,4,6,8 HSRP priority higher on Core 1. For VLAN 3,5,7,13 HSRP priority higher on Core 2.

If you could provide a topology, i can suggest you on your uplinkfast setup.

Hope this helps.

Thanks

Vivek

[Sihanu N]

Hi Vivek,

Many Thanks for the detailed reply to my post.

We had two core switches and 4 distribution switches and around 26 access switches in our network. The core swithches was working like HSRP active/standby mode, but recently the core 1 fails due to some hardware issues. Now core 2 is active and after that some printers in vlan 1 showing some slow perfomance (some times about 5 to 10 minutes) while printing from other vlans other(within the vlan there is no performance issues). Anyway it will be fixed and we are going to load balance the HSRP once the core 1 hardware is fixed. Now we are looking to load balance the STP for the same.

Following are my findings

1) Spanning-tree root bridge is an access switch for all vlans (spanning-tree mode - pvst)

2) Some initial ping loss is to printers from machines experiencing print issue

3) CPU/memory utilization of all network devices in the environment shown normal even after core 1 failure

4) Cost to root bridge from core switch for all vlans are 3008

My Queries

1) Is the issue related to wrong spanning-tree root bridge election? If then how can we find it?

2) Only the priority change in core switch required for the spanning-tree load balancing between core swithces?

Thanks & Regards,

Sihanu N

[Vivek Ganapathi]

Hi Sihanu,

I feel this to be an issue with wrong STP root bridge election. But need further troubleshooting. How i troubleshoot a Layer 2 issue

1) I draw an entire topology & write up which one is the Root bridge, designated port, root port & the ones blocked. But in your case, the spanning-tree was left with default settings which means, it took up the one with the least MAC-address switch in the network (in other way, an oldest switch ). To overcome this, being an admin you should manually tune this configuration by configuring the STP priority. As in your case, it's left to default start troubleshooting from your CORE-2

• Find the root ports.
  
• Trace the path to the root bridge
  

2) Yes, by doing the priority change you ensure that you have manually decided which switch would be the root bridge for specific VLANs. The one with the lowest priority will be elected as the root bridge. So to effectively load balance between your core switches, you need to plan it. You can use my suggestion. Also, there is one more option & many would like it. Rather than configuring a priority value, you can just say it to be a root primary or secondary for those VLANs

CORE-1

spanning-tree vlan 2, 4, 6, 8 root primary

spanning-tree vlan 3, 5,7,13 root secondary

CORE-2

spanning-tree vlan 3, 5,7,13 root primary

spanning-tree vlan 2, 4, 6, 8 root secondary

NOTE: Most Important !! Do this change if you are planning to, only during non-production hours. Else, you may end up in an outage.

Hope this helps.