Junos OS

last person joined: 19 hours ago 

Ask questions and share experiences about Junos OS.
  • 1.  J6350 ES Cluster/HA

    Posted 01-24-2009 16:33

    I need some help setting up 2 J6350's into an Active/Passive cluster.  I am very familar with NSRP on the SSG devices but am having a hard time grasping the concepts of how to do this in JunOS.

     

    From what I've read in the security doc for the J6350 this is acomplished using clustering with redunancy groups.  In particular I'm curious about how to configure IP Addresses on the 2 devices.  How does JunOS handle the creation of a VIP to be shared between the two? 

     

    In NSRP you would configure the same IP on the interfaces with different mgmt ip's and then setup NSRP and it would create a VIP for that interface using the IP configured on that interface.  How do you do this in JunOS?



  • 2.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 05:45

    Have you seen this article? http://kb.juniper.net/KB12397

     

    It's specifically for JUNOS with Enhanced Services so if you're running JUNOS (router mode) I don't think it applies.

     

    Regards,

     

    -Keith

     

     



  • 3.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 07:06
    how do i see what physical interface it maps to fxp1?


  • 4.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 07:40

    and the J series requires a different interface for the data link correct? I know on the SSG using NSRP both control and data links can be the same interface.

     

    edit:

     

    Also, instead of blowing an interface for mgmt can i assign different ip addresses to the physical interfaces that are members in a reth?

     

    ie

     

    reth0 members = ge-0/0/0 and ge-7/0/0 (one on both devices in the cluster

    reth0 ip = 192.168.0.1

    ge-0/0/0 = 192.168.0.2

    ge-7/0/0 = 192.168.0.3

     

    ip addy on gig interfaces would be to manage each device?

    Message Edited by brmurphy on 01-26-2009 08:08 AM


  • 5.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 08:11

    The fabric link is the data link equivalent. Both control and data link need to be different for JUNOS with enhanced services, they cannot be the same link.

     

    Interface                       

     

    ge-0/0/2        fxp0  - Management  

    ge-0/0/3        fxp1  - Control link

    fab0                       - Data link on node 0

    fab1                       - Data link on node 1



  • 6.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 08:24
    how can i verify that the fabric links are functioning correctly?  when i do a show interfaces on fab0 and fab1 they show down althought the cable is plugged in.


  • 7.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 08:32

    The working of JSRP is different from NSRP. In JSRP both the unit in a cluster work as one device. The RE will be active on say node0 and the interface which is part of RG can be active on either node0 or node 1. In case of JSRP only the primary interface in the RG group will be active and accept traffic. This is not possible in NSRP, in NSRP the VSI interface has to be active on the unit which is primary.

     

    The fxp0, fxp1 interface are reassigned and renamed when chassis cluster is enabled and fabric i.e. fab0 and fab1 needs to be configured.  Configuring an IP address on the interface which is part of reth0 will not work. 

     

    Unfortunately you can't reuse the fxp0 interface as a reth interface. Hope this helps. 🙂



  • 8.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 08:45
    to what interfaces are fxp0 and fxp1 reassigned?


  • 9.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 09:20

    Sorry for the confusion.

     

    I meant ge-0/0/2 and ge-0/0/3 interfaces are reassigned as fxp0 and fxp1 respectively when chassis cluster is enabled.



  • 10.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 09:27

    ok, just to make sure i'm understanding this correctly.

     

    so i probably shouldn't make fab0 and fab1 ge-0/0/2 and ge-7/0/2 if those interfaces are being reassigned to fxp0 and fxp1 for management.

     

    This is because I'm forced to use one interface per device for mgmt only, that interface is ge-0/0/2 and ge-7/0/2 which gets reassigned to fxp0  and fxp1  when clusting is enbled.

     

    so i need to make any inferface asside from ge-0/0/2 and ge-7/0/2 my fab0 and fab1 links.



  • 11.  RE: J6350 ES Cluster/HA
    Best Answer

    Posted 01-26-2009 10:19

    The below is how the chassis cluster needs to be physically connected.


     Node 0                                                                                         Node1

    ge-0/0/2 =fxp0                                                                          fxp0    ge-0/02 
    ge-0/0/3 =fxp1  - Control link                          -------------------- fxp1    ge-0/0/3
    <Any-interface>= fab0- Data link on node 0  --------------------  fab1   <Any-interface> Data link on node 1

    Connect fxp0 to your network switch for mgmt.
    fxp1 is connected back to back between the 2 routers.
    fab0 and fab1 is connected back to back between the 2 routers.

     

    fab0 interfaces will be any interface from 0/0/x to 6/0/x except ge-0/0/2 and ge-0/0/3 

    fab1 interfaces  will be any interface from 7/0/x to 13/0/x except ge-7/0/2 and ge-7/0/3

     



  • 12.  RE: J6350 ES Cluster/HA

    Posted 01-26-2009 07:29

    The control link on J-Series is ge-0/0/3. When you enable chassis cluster the ge-0/0/3 interface is renamed as fxp1.

     

    To set up the control link on services routers, you connect the control interfaces on the two devices back-to-back. When you initialize a services router in cluster mode, JUNOS software with enhanced services renames the control interface to fxp1 and uses that interface for the cluster control link. To enable the control link to transmit data, the system provides each fxp1 control link interface with an internal IP address. 

     

    MR