Comware

 View Only
  • 1.  MAD LACP on intermediate 5700 IRF switch

    Posted Feb 15, 2023 08:51 PM

    Hi All,

    I have a current setup of the following:

    2-member Comware 5130 IRF (connected via BAGG28 to upstream switch) > 6-member 5700 IRF (connected downstream via BAGG5 to the 5130 and upstream via BAGG10 to VSF) > 2-member 5406R Core VSF (connected downstream via Trk10 to the 5700s)

    I have mad enable on the 5130 BAGG28, i have mad enable on the 5700 on both BAGG5 and 10, and on the VSF have lacp mad-passthrough enable on the Trk10

    When i go into the most downstream switch, the 5130 IRF, running the command display mad verbose it states the MAD status is normal and member ID 1 and 2 normal

    When i log into the most upstream switch, the Core VSF, running the command show lacp mad-passthrough (counters) i can see Trk10 enabled and PDU for Rx/Tx

    When i log into the intermediate switch, the 5700 IRF, running the command display mad verbose it states the MAD status for both BAGGs is faulty.

    On the intermediate switch i initially only had mad enable on the BAGG10 upstream link to VSF, but that stated it was faulty, so i thought i might need it enabled on both the uplink/downlink BAGGS, but it states both are faulty.

    The 2 connecting IRFs have different domains (0, 50), and VSF has domain 1

    Note: i recently extended the 5700 from a 4-member ring to a 6-member ring but have only configured the first 4 members in the BAGG to the upstream VSF. 

    Any help in resolving this intermediate switch MAD status issue would be appreciated. Configs below:

    Downstream switch (5130 IRF):

    <5130-IRF>display mad verbose
    Multi-active recovery state: No
    Excluded ports (user-configured):
    Excluded ports (system-configured):
      IRF physical interfaces:
        Ten-GigabitEthernet1/0/25
        Ten-GigabitEthernet2/0/25
    MAD ARP disabled.
    MAD ND disabled.
    MAD LACP enabled interface: Bridge-Aggregation28
      MAD status                : Normal
      Member ID    Port                                  MAD status
      1            Ten-GigabitEthernet1/0/28             Normal
      2            Ten-GigabitEthernet2/0/28             Normal
    MAD BFD disabled.
    <5130-IRF>

    Intermediate switch (5700 IRF):

    <5700-IRF> display mad verbose
    Multi-active recovery state: No
    Excluded ports (user-configured):
    Excluded ports (system-configured):
      FortyGigE1/0/41
      FortyGigE1/0/42
      FortyGigE2/0/41
      FortyGigE2/0/42
      FortyGigE3/0/41
      FortyGigE3/0/42
      FortyGigE4/0/41
      FortyGigE4/0/42
      FortyGigE5/0/41
      FortyGigE5/0/42
      FortyGigE6/0/41
      FortyGigE6/0/42
    MAD ARP disabled.
    MAD ND disabled.
    MAD LACP enabled interface: Bridge-Aggregation5
      MAD status                : Faulty
      Member ID    Port                                    MAD status
      1            Ten-GigabitEthernet1/0/38               Normal
      3            Ten-GigabitEthernet3/0/38               Normal
    MAD LACP enabled interface: Bridge-Aggregation10
      MAD status                : Faulty
      Member ID    Port                                    MAD status
      1            Ten-GigabitEthernet1/0/32               Normal
      2            Ten-GigabitEthernet2/0/32               Normal
      3            Ten-GigabitEthernet3/0/32               Normal
      4            Ten-GigabitEthernet4/0/32               Normal
    MAD BFD disabled.
    <5700-IRF>

    Upstream switch (5406 VSF):

    Core-VSF#    sh lacp mad-passthrough

      Trunk-Group  LACP-MAD-PASSTHROUGH
      ------------ ---------------------
      Trk10        Enabled
    Core-VSF#    sh lacp mad-passthrough  counters

                    MAD Passthrough  MAD Passthrough  MAD Passthrough
      Port   Trunk  PDUs Tx          PDUs Rx          PDUs Dropped
      ------ ------ ---------------- ---------------- ----------------
      1/C7   Trk10  2747945          915987           0
      1/C8   Trk10  2747939          915993           0
      2/C7   Trk10  2747956          915976           0
      2/C8   Trk10  2747956          915976           0

    Regards,

    James



  • 2.  RE: MAD LACP on intermediate 5700 IRF switch

    Posted Feb 16, 2023 11:04 AM

    Hi! so who is the intermediate entity that is acting as MAD for what stacks?

    Since you have three stacks (two IRF and one VSF) and they are chained together in this way:

    Top VSF 2x5400R zl2 <--> Central IRF 6x5710 <--> Bottom IRF 2x5310

    It seems you're using the VSF as the top MAD for the IRF 6x5710 and the central IRF 6x5710 as the MAD for the bottom IRF 2x5310...I would have expected (physical topology and cabling permitting) that one intermediate entity only was used as the intermediate switch to deploy as the MAD against multiple stacks directly connected to that switch.

    Or could be the vice-versa where the central IRF 6x5710 is acting as the intermediate device for MAD for both top VSF and bottom IRF 2x5310...

    MAD physical topology (as an additional layer to the current physical topology):

    Top VSF 2x5400R zl2 <-- LACP Port Trunk n -- MAD BAGG x --> Central IRF 6x5710

    Top VSF 2x5400R zl2 <-- LACP Port Trunk m -- MAD BAGG y --> Bottom IRF 2x5310

    But maybe I'm wrong...




  • 3.  RE: MAD LACP on intermediate 5700 IRF switch

    Posted Feb 16, 2023 11:12 PM

    Hi Parnassus,

    Thank you for those suggestions mentioned below. I may need to clarify, as i think i have it setup wrong.

    The top Core VSF switch will eventually be configured to use OOBM-MAD to a seperate OOBM switch, for MAD and management of itself.

    I was then hoping the top VSF switch could then act as the switch to manage the MAD status for both the downstream IRFs (the intermediate 5700 IRF and the bottom 5130 IRF), even though there is only a single (trunked) physical path from the bottom IRF (5130) to the intermediate IRF (5700) to the VSF (5406).

    But maybe, as per your suggestion in the MAD physical topology below, if the Core VSF will be the central MAD detection switch, any IRF switch that has MAD enabled has to have a direct physical connection to the VSF.

    I have also tried removing the mad enable from BAGG10 and lacp mad-passthough enable from Trk10 (connection between intermediate 5700 and top 5406), thus just having the mad enable between the 2 IRF switches, with each IRF switch using each other to detect the MAD status, and i get the same result, the bottom 5130 IRF MAD status is normal, but the 5700 IRF MAD status is reporting faulty.

    Food for though > could this due to the fact i have a 6 member 5700 IRF ring, but an using only member 1 and 3 for the BAGG5 down to the 5130. 

    Any other suggestions would be appreciated.

    Regards,

    James




  • 4.  RE: MAD LACP on intermediate 5700 IRF switch
    Best Answer

    Posted Feb 19, 2023 05:08 PM

    Hi James, implementing LACP MAD requires that each IRF member of a IRF must have a link with the intermediate device...it means that your 6 members IRF should require a 6 members links (MAD enabled) LACP to the upstream VSF and your just 2 members LACP looks not enough to obtain a fully effective MAD.




  • 5.  RE: MAD LACP on intermediate 5700 IRF switch

    Posted Feb 23, 2023 11:05 PM

    Hi Parnassus,

    As per your suggestion above, managed to get MAD LACP working on the intermediate 6-member IRF. Needed to increase the trunk size from 4 x 10Gb to the top VSF, to 6 x 10Gb (most important part was utilising a link from each 5700 member, M1 through to M6, of the 5700 IRF).

    Thus, final configuration had the bottom 2 x 5130 IRF using mad enable on BAGG28 uplink to intermediate 6 x 5700 IRF, then on the 5700s, used mad enable on the 6 x 10Gb BAGG10 trunk uplink (1 link from each member) to the 2 x 5406R VSFs. On the top Core VSF, had lacp mad-passthrough enable on the downlink Trk10 down to the 5700's

    In essence, this what what i was hoping to achieve, the downstream distribution and access layer 5700/5130s utilising the Core VSFs for MAD monitoring of the comware switches. Once again, thanks for your insights.

    Regards,

    James