ISIS infinite metric

I have two XR routers that are ISIS L2 peers with each other over a WAN link.  REcently the link had a fiber outage and the link flapped a few times and was down about 8 hours or so...

 

When the link came back ISIS neighborship came back up ok, and IPv4 recovered, but in IPv6 I see they have an "infinite" metric on each others learned routes:

 

show isis ipv6 route:

L2 2001:550:2:17::3:0/112 [<infinity>/115]
L2 2001:1900:2100::720/126 [<infinity>/115]

 

When there is an infinity metric, the link is never used.  When I look at the links for IPv6 on each side, it's telling me the link metric is the detault of 10.  Any idea what causes this and how to fix it?  Resetting the ISIS process did not seem to do the trick.

-v10d

 

Comments

  • I found my answer, which seems like some kind of bug.  When the link recovered one side of the connection showed up as if it had no IPv6 link-local address; so I think the routing protocol was exchanging but there was no valid next-hop; therefore making it "infinite".  Strange.  The only way to fix it was to remove the ipv6 address on the effected side commit and re-commit it back.

     

      IPv4 Unicast Topology:    Enabled
    Adjacency Formation: Running
    Prefix Advertisement: Running
    Metric (L1/L2): 10/10
    MPLS LDP Sync (L1/L2): Disabled/Disabled
    IPv6 Unicast Topology: Enabled
    Adjacency Formation: Disabled (Not up in intf protocol stack)
    Prefix Advertisement: Disabled (Not up in intf protocol stack)
    Metric (L1/L2): 10/10
    MPLS LDP Sync (L1/L2): Disabled/Disabled

    IPv4 Address Family: Enabled
    Protocol State: Up
    Forwarding Address(es): x.x.x.x
    Global Prefix(es): x.x.x.x/30
    IPv6 Address Family: Enabled
    Protocol State: Down (IPv6 not up in intf protocol stack)
    Forwarding Address(es): Unknown (No link-local address exists)
    Global Prefix(es): aaaa:bbbb:b:201::a/127
  • This was actually a bug I hit during the SP lab. I had v6 on both sides of the links, but I could not ping from directly connected to directly connected v6 addresses (v4 worked fine). This was on the 7200 IOU devices. Its not that the ping would fail in the data-plane, but when I tried to ping it would say 'no valid local source address to source from'.

    I also encountered this same issue while doing the Gold labs, and the way that I solved it was just to reboot the 7200 instance (15 seconds to come back up). 

     

  • I've seen something similar to this in production on both the 7200 VXR NPE-G2 platform and on the CRS-3.  Every time the problem was resolved by removing and readding the v6 address.  If i recall correctly i could ping the assigned v6 address but couldn't ping the link-local address.

Sign In or Register to comment.