Morning everyone .. i am looking to get some form of help from the community about a frustrating sets of problems we are having with MLAG and 4 M7100s switches .I would hopefully receive back a running and effective MLAG config from someone with active MLAG
to be inspired by and find which steps we are missing .
Making a long story short this is the issue which is troubling my sleeps ...
We have 4 M7100 switches (sw1 - sw 2 - sw3 - sw4) (latest firmware)
and another one (a M7100 as well but in use as a simulation of a 1gb switch)
In the image enclosed there are the roughl connections
in a few words we want to use the sw1-sw4 as a switch stack and with an MLAG from this MLAG capable stack to a non MLAG capable switch achieve to objective to get rid of spanning tree
We have a lag between sw1 and sw3 and between sw2 and sw4
then we have a peer lag between sw1 and sw2 (as per the MLAG manual pages)
and an MLAG between one sw1 and sw2 port to a 1gb stack switch in the same set up in the same LAG
The 1gb switch is not MLAG capable and thus is setup as a 2 port LAG of the same MLAG LAG number
the vlan 1 gateway 10.30.0.1 is active on a ubuntu box connected to the labswitch
here follows the configurations ...
what we would like to achieve is having a connection between hosts connected in the 10G switch stack with hosts connected in 1G switch (in the same private B class) with or without routing ....
here follows the configs (lab switch is known to be not complete (probably here there are the missing points ..) Unfortunately afaik there is no existing complete running config which can help us except some quite obscure and short manual pages.
actually from either switches interfaces nor hosts connected to these interfaces can access
counterparts in the labswitch.Hosts connected in the 10G stack are visible each other
SW1 hostname "sw1-10g-eshelter" vlan database vlan routing 1 1 exit configure line console session-timeout 60 exit line telnet exit interface 0/2 addport lag 1 exit interface 0/3 addport lag 1 exit interface 0/23 addport lag 2 exit interface 0/24 addport lag 2 exit interface 0/1 addport lag 3 exit snmp-server sysname "sw1-10g-eshelter" ! udld enable
interface 0/2 udld enable exit interface 0/3 udld enable exit interface lag 1 no spanning-tree port mode switchport mode trunk vpc peer-link exit interface lag 2 switchport mode trunk exit interface lag 3 vpc 1 exit interface vlan 1 routing ip address 10.30.0.11 255.255.0.0 exit ip default-gateway 10.30.0.1 feature vpc vpc domain 1 peer-keepalive enable exit exit
I inquired your concern to the higher tier of NETGEAR Support. As per the higher tier of NETGEAR Support, the configuration you have posted is correct. Let me share the article below:
Please note what is says about Redundancy from the article above: “When connecting equipment to MLAG switches, this connection should be done redundantly. Ideally, non-redundant equipment is connected through a redundantly-connected edge or core switch.”
what we would like to achieve is having a connection between hosts connected in the 10G switch stack with hosts connected in 1G switch (in the same private B class) with or without routing ....
On SW1 and SW2, you can only connect devices by using LAGs. Therefore, a PC with a single NIC card that is connected to either SW1 or SW2, will not be able to communicate with Lab Switch.
Kindly answer the questiions below:
a. On SW1 and SW2 connectivity needs to be achieved by the use of LAGs. Therefore, if you connect a host without configuring a LAG, the host will not be able to connect to Lab Switch. Are LAGs used for every device connected to SW1 or SW2?
b. The expected behavior is that on SW3 and SW4, no LAGs are needed to be configured on the connected devices in order for them to be able to have full connectivity with the Lab Switch. Can you confirm that this is the actual behavior?
On SW1 and SW2, you can only connect devices by using LAGs. Therefore, a PC with a single NIC card that is connected to either SW1 or SW2, will not be able to communicate with Lab Switch if the single NIC PC isn't connecting to an MLAG. This is a limitation and we know this is "consuming" one port on the other switch that will never be used. But this is the only solution. Please create one MLAG across SW1 and SW2, and connect the "orphan" PC to SW1 this way. Please create another MLAG to connect another "orphan "PC" to SW2.
Etc.
This is not elegant, that's why we enhanced our Stacking technology in more recent generation switches (M4300 series) for allowing spine and leaf stacking between 1G and 10G models, distributed LAG across the stack with Non-Stop Forwarding (LACP), etc.
Just want to follow-up on this. Let us know if you have further concerns.
Otherwise, if ever your concern has been addressed or resolved, I encourage you to mark the appropriate reply as the “Accepted Solution” so others can be confident in benefiting from the solution. The NETGEAR Community looks forward to hearing from you and being a helpful resource in the future!
Sorry for my delayed answer i was off for a few days (to recover after the fight with the Netgear switches )
Coming to your questions and notes ...
I read the articles so many times i can repeat it while on waterboarding ) however it did not help .The best should be to have a complete running config .I requested support and i received it .. actually i did not have the time to test it but will do and share the results
a: I configured a server with two netowrk cards in a bonding configuration with a leg in sw1 and a leg on sw2 ... The result was not as planned (communicating with the lab switch)
b: i tried as well to connect a single network card host on sw3 with the same depressing results as above
As per your understanding an host connected to the lab switch could be single carded isn't it ?
The same for sw3 and sw4 while on sw1-2 the connection has to be made on each side in bonding ( wondering if the bonding type can make a difference)
Then what happened was a complete debacle except some success pinging the lab switch from a correctly connected double carded host on sw1-2 but was not possible to ping an host connected to the lab switch in single connection ...
this is what i have to share right now .. hopefully this week beginning of the next i would try the configuration Netgear support gave me
However i am aware and accepted from the begginning the limitations and were planning to connect the hosts to any of sw1-2-3-4 switches always with the connection (sw1 and 2 or sw3 and 4)
the goal was to communicate with an (single or double carded) host connected to the lab switch .Wtih the given configs I tried all the matrix (single and bonded connection on both sides)
with no success except connectiong from the sw1-2 mlag to the labswitch ip
In the meantime i opened a support case with the support and received a configuration from your lab
I have to test it yet and hope to do it before the end ./ beginning of the next week and will share the results
)
) however it did not help .The best should be to have a complete running config .I requested support and i received it .. actually i did not have the time to test it but will do and share the results
