Forum Discussion

Renato's avatar
Renato
Icon for Altostratus rankAltostratus
Apr 23, 2015

Disabling data center in GTM configuration and config synchronization

Hi, experts!

I have the following GTM configuration:

DatacenterA    =>  ServerA  =>  Virtual Server: 1.1.1.1
DatacenterB    =>  ServerB  =>  Virtual Server: 2.2.2.2
WIP: test.com  =>  WIPPool  =>  1.1.1.1, 2.2.2.2

In normal conditions all queries for test.com return both the addresses, what looks OK. The problem is that we have to disable one data center that will be used only in planned situations. I'm disabling one of them for testing, but GTM is still returning both the addresses. Is there a specific parameter or whatever to get this working?

Another thing I noted is that the configuration is not being replicated. I've enabled the GLSB Synchronization to group default and executed all the procedure, even the 'tmsh run gtm gtm_add' command that copied the initial configuration, but always when I want to replicate the configuration I have to execute the gtm_add script again. I suppose it should work automatically like a config-sync group with Automatic Sync enabled, shouldn't it?

I would appreciate any help.

Thank-you,

Renato A. Ferreira

2 Replies

  • GTM is still returning both the addresses?

     

    Seems you have selected Round Robin Method. So traffic is going both Pool, even though disable. In order to achieve this task.. Select Method as Global availability. Hope problem will resolved.

     

    GTM is not syncing automatically?

     

    It sounds like your standby unit may not be trusting your primary. Under System -> Configuration -> Global Traffic -> General, is your synchronization group name the same on both units?

     

    Also, when you perform a 'netstat -an' on your standby unit, is it listening on 4353?

     

  • I managed as following to get everything working here:

     

    1. First of all, once I started to use a self-ip address as listener and used this address as server address, the GTM has started to recognize itself and the others devices and since then the synchronization group, discovery e everything started to work as the related documentation suggests.

       

    2. As I was using IP address in WIP pools without direct relationship with local virtual servers, I just created a fake monitor to make all objects up without really checking them.

       

    Doing it, the DNS queries started to be really balanced in a round-robin fashion instead of returning always both addresses and, finally, stopped to return the address from a disabled data center.

     

    I'm only worrying about having to leave the iQuery port open for a public accessible DNS address in the port lockdown settings for all self-ips used as listener. It's still sounding weird for me.