Multicast problem between X670G2-48x-4q and HP ProCurve 5404 zl - Sporadically no Multicast

[André_Struwe]

Dear Community Members,

I am very frustrated and do not know how to find a solution for my problem. The following technical environment currently is in use:

• 2 x X670G2-48x-4q stacks (22.4.1.4), connected via ISC as core with IGMP snooping enabled on both switches.
• 8 x HP ProCurve 5404 zl switches as distribution layer, connected via mlag to both X670G2-48x-4q stacks and IGMP snooping enabled (ProCurve uses a trunk port for link aggregation)
• one IGMP querier connected to one of the X670G2-48x-4q core switches
• no multicast routing is done because every sender and receiver is connected within the same VLAN
• IGMP V2 on all switches
• configuration straight forward/out of the box on all switches with 4 VLANs where IGMP is enabled

Here comes the problem:

If I connect a multicast sender to the HP ProCurve and a receiver to the the first or second X640 core stack, multicast sometimes works and sometimes not.

The following has been tested so far:

• connecting MC sender and receiver to one X670: MC is working fine
• connecting MC sender and receiver across both X670 stacks (with ISC in the middle): MC is working fine
• connecting MC sender/receiver to HP ProCurve 5404 zl and sender/receiver to both X670 stacks: MC sometimes works and sometimes not

Both stacks and the HP ProCurve do form a local multicast group via IGMP snooping with the correct receiver port which has been snooped by IGMP reports from the receiver.

The HP ProCurve tells me that he sees a querier on his downlink and should send IGMP reports through that port back to the querier across the core stack. Even the core stack tells me that there is a querier on the lan and I can assure that there is only one querier present.

But it seems that sometimes either the X670 or the HP ProCurve does not see the receiver on it’s partner (which it should learn with the help of IGMP reports or queries). So, IGMP snooping table does only make an entry for the member port of the receiver, not the uplink.

What is really weird:

Sometimes multicast stops after some time. Then I switch the receiver from core 1 to core 2 and multicast is working again.

Has anybody ever seen such a behaviour or does know how to solve the issue?

Any help is appreciated .

Thanks in advance

Below is a summary of one of the X640G2 configs. Be aware that I’ve shortened the config because there are more VLANs and ports but that should be the only thing that might be necessary.

Slot-1 X670-Core1 # show configuration
#
# Module devmgr configuration.
#
configure snmp sysName "X670"
configure timezone name MEZ 60 autodst name MESZ begins every last sunday march at 2 0 ends every last sunday october at 3 0
configure slot 1 module X670G2-48x-4q
configure sys-recovery-level slot 1 reset
configure slot 2 module X670G2-48x-4q
configure sys-recovery-level slot 2 reset


#
# Module vlan configuration.
#

enable jumbo-frame ports all
configure vlan untagged-ports auto-move off
create vlan "ISC"
configure vlan ISC tag 4000


enable sharing 1:1 grouping 1:1,2:1 algorithm address-based L2 lacp
enable sharing 1:7 grouping 1:7,2:7 algorithm address-based L2 lacp

configure vlan Default add ports 1:1 untagged
configure vlan ISC add ports 1:1 tagged


#
configure cdp device-id system-mac
enable cdp ports 1:1
enable cdp ports 1:2



#
# Module hal configuration.
#
configure iproute sharing max-gateways 4

#


#
# Module ospf configuration.
#
configure ospf routerid 123.345.678.123
enable ospf
configure ospf add vlan Default area 

#
# Module ospfv3 configuration.
#
configure ospfv3 spf-hold-time 10

#
# Module stp configuration.
#
create stpd DEMO
configure stpd DEMO mode dot1w
configure stpd DEMO priority 0
configure stpd DEMO default-encapsulation dot1d
configure stpd s0 mode dot1w
configure stpd s0 default-encapsulation pvst-plus
enable stpd DEMO auto-bind vlan Default
enable stpd DEMO auto-bind vlan ISC
enable stpd DEMO auto-bind vlan VLAN_
enable stpd DEMO auto-bind vlan VLAN_

configure stpd DEMO ports mode dot1d 1:1
enable stpd DEMO
disable stpd s0


#
# Module vrrp configuration.
#
create vrrp vlan Default vrid 1

#
# Module vsm configuration.
#
create mlag peer "X670-Core2"
configure mlag peer "X670-Core2" ipaddress 169.254.0.2 vr VR-Default
enable mlag port 1:3 peer "X670-Core2" id 3
enable mlag port 1:4 peer "X670-Core2" id 4

[André_Struwe]

Hello Nikoly,

thank you for your reply.

We found out that in order to solve the problem we had to do a restart of all HP ProCurves connected to the X670.

Nethertheless a software update of X670 helped us to create a new MLAG for a totally new HP IRF cluster which couldn’t be connected properly to the X670 through MLAG.

So, in the end a reboot of the HP ProCurves has solved the Multicast/IGMP issue.

Thx

[Necheporenko__N]

Hello André,

The above config is incorrect. The data carrying VLANs that are added in the uplink port of the downstream switch should be added on both the MLAG ports and the ISC ports of the peer switches.

For more details please check https://gtacknowledge.extremenetworks.com/articles/How_To/How-to-configure-MLAG-in-Extreme-switches/... 

Best Regards,

Nikolay