Wired Intelligent Edge

Hi, we have a full Procurve infra-structure. Our 2 main server switches are Procurves 2810-48G and our edges switces are the 2610-48G-PWR.

We recently bought an IBM Blade Center and it came with Cisco 3012 switches.

We made 4 different LACP port Trunks out of the 4 external interfaces of the Ciscos towards our 2 main 2810 Procurves.

That gives us 4Gig Trunks out of each ciscos.

For some reasons this ALMOST works. For some reasons 2 of the 4ports Trunks are going down.

I don't have much logging in the Procurves about this issue, only the following when it occures:

I 09/22/10 19:13:01 ports: port 17 in Trk1 is now off-line
I 09/22/10 19:13:01 ports: port 18 in Trk1 is now off-line
I 09/22/10 19:13:01 ports: port 35 in Trk1 is now off-line
I 09/22/10 19:13:01 ports: trunk Trk1 is now inactive
I 09/22/10 19:13:01 ports: port 36 in Trk1 is now off-line


The Ciscos side is a bit more chatty...

- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Gi0/15, putting Gi0/15 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Gi0/16, putting Gi0/16 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Gi0/17, putting Gi0/17 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Gi0/18, putting Gi0/18 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/15 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/16 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/17 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/18 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Po1 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: lsgroup error detected on Gi0/1, putting Gi0/1 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: lsgroup error detected on Gi0/3, putting Gi0/3 in err-disable state
- 1d05h: %PM-4-ERR_DISABLE: lsgroup error detected on Gi0/4, putting Gi0/4 in err-disable state

I just don't know where to start looking has far as what is our ROOT cause issue.

If I go to the cisco and do a shut/no shut on the po1 (Trunk) interface...the link comes back on-line for many hours... and it eventualy goes down again.

Would could cause this to be so unstable?


Regards,


Andy

 

 

P.S. This thread has been moved from Switches, Hubs, Modems (Legacy ITRC forum) to ProCurve / ProVision-Based. - Hp Forum Moderator

Configuration of STP on your Cisco switch? I think this is PVST. Procurve by default uses MSTP. Try to bring the configuration of STP to the same. Here are some helpful links on the topic:

http://cdn.procurve.com/training/Manuals/ProCurve-and-Cisco-STP-Interoperability.pdf

http://www.tecnocael.it/ftp/docs/ProCurve_Cisco.pdf

http://www.g33k.de/blog/wp-content/hp-cisco%20trunking.pdf

Thanks for the link,


I run the following STP on the procurves:

spanning-tree priority 0 force-version RSTP-Operation


The Ciscos do run the following:

spanning-tree mode rapid-pvst



Am I to understand that these are 2 STP versions completely different that are not compatible with each others?



Thanks


Andy

Andy,

Did you put the external interfaces of the Cisco bladeswitch in one 4-port trunk (link aggregate) and you terminate two of the 4 ports on one switch and two on another switch? If that is the case, that will not work unless you use some kind of vendor specific implementation. A trunk (link aggregate) has to be terminated on the same switch unless you use propriatory technology. ProCurve (HPN) has something called "distributed trunking" on their higher end switches. The products you mentioned do not support this. Do you have a topology drawing?

Hi,

The Trunks are terminating on the same Procurve switches. I am not trying to terminate the bundled ports on different destinations.

I provided a drawing.

yes, two different versions of spanning tree.
PVST sends a BPDU for every VLAN, RSTP just sends one BPDU (besides other differences).
You need to find a common version for both, Cisco blade switch and ProCurve. I don't know what different versions like MST, STP, RSTP Cisco supports on their bladeswitch.

if you are using link aggregation then there is no need for STP, what you have here is that the procurve switches forward the RSTP packets since they don't know the Cisco RSTP multicast address and consider and forward it to all ports and the Cisco ports then receive the same packets again, not 100% sure, but i think there is a command that might help you on this.

[no] spanning-tree [ETHERNET] PORT-LIST pvst-filter

Stop a specific port or ports from receiving and retransmitting PVST BPDUs. The command
indicates
which ports are not expected to receive any PVST BPDUs.
Default: Disabled on all ports.

So try to enable this on the links connected to the Cisco switch and see what will happen.
the command will be

spanning-tree <PORT list=""> pvst-filter</PORT>

Hi Shadow,

Thanks for the details you provided. Just a side note, I could not use the

spanning-tree trk1 pvst-filter

because it is not there on the 2810 but I used instead the

spanning-tree trk1 bpdu-filter

and since then, the LACP link has been up for 24 hours now on the link

I will go on and add this command to the 3 other TRK links and I will update again tomorrow.

Thanks!

TRUNKs are still UP. Looks like it stabilized with the bpdu-filter commands in place.

Now one thing I was wondering is, if on Blade, I am to Team some NICs in a Active/STandby fashion... would that create network loops?

do you have a copy of your HP and cisco configs? I am looking to do something similar on my network as i have a cisco 3850 stack that i want to connect my HP2530 switches to with dual links.

After close to a month of uptime without any LACP link going down, one server had to reboot for maintenance update of the OS. Upon reboot, one of the LACP Trunk went down.
I re-enabled it manualy.

I am wondering if this could be caused because only the Procurve side is filtering the BPDUs. SHould I consider disabling BPDUs on the Cisco side of the LACP TRUNKs also?