Forum Discussion

PiotrL's avatar
PiotrL
Icon for Cirrus rankCirrus
Jun 23, 2015

GTM - offers two addresses instead of one

Hi, we have two locations (two ltm pairs, and two GTMs). Normally GTMs offer one address (wide ip) of the VS in the primary location (global availability option). When VS in the primary location goes down GTMs start serving two addresses (from primary location and secondary location) instead of only one (working in secondary location). The log shows: notice tmm1[15147]: 011ae025:5: UNRESOLVED rrtype = A, wip = some.name, LDNS(67.222.132.211), Return to DNS

 

So why "return to DNS" ?

 

availability
BIG-IP DNS
deployment

4 Replies

Sort By
  • TechT_163800's avatar
    TechT_163800
    Icon for Nimbostratus rankNimbostratus
    Jun 23, 2015

    Hi PiotrL, Is the primary and secondary VS configured under same pool ? If so, can you please check is there any load balancing method configured for "Alternate" and "Fallback" under WideIP pool configuration instead of Global Availability ?

     

  • Stanislas_Piro2's avatar
    Stanislas_Piro2
    Icon for Cumulonimbus rankCumulonimbus
    Jun 23, 2015

    GTM return as many answer as configured in GTM pool. Default value is 1.

     

    If preferred and Alternate load balancing method is not exhaustive (topology with not all IPs defined, RTT for an IP not probed or blocking request, ...) or all pool members unavailable, the fallback is return to DNS by default.

     

    When creating wideIP, zone runner add wideIPs to BIND configuration. The default behavior of BIND is to answer with all records of wideIPs.

     

  • PiotrL's avatar
    PiotrL
    Icon for Cirrus rankCirrus
    Jun 23, 2015

    There are two different pools for Wide IP on GTMs (pool-vs-primary and pool-vs-secondary). Load balancing method for Wide IP: global availability. Load balancing for a pool: preffered & alternate is round robin, fallback: return to dns.

     

  • TechT_163800's avatar
    TechT_163800
    Icon for Nimbostratus rankNimbostratus
    Jun 24, 2015

    Hi PiotrL, i suggest you to configure both the members under one pool and mention the "Preferred LB Method = Global Availability" inside Pool (not for the Pool). Hopefully this will resolve your issue.