Forum Discussion

SynACk_128568's avatar
SynACk_128568
Icon for Cirrostratus rankCirrostratus
Mar 09, 2015

GTM sync group working

Hi All,

 

wanted to confirm working of gtm in SYNC group :

 

If there is a sync group containg 5 GTM's in 1 asia ,3 US and 1 europe . And the GTM has servers configured and using bigip monitor iquery to report status to GTM for name resolution .

 

If in case one of the GTM marks virtual servers down hosted on LTM due to issue in iquery .

 

1) Will this be propagated to other GTM's ?

 

2) suppose GTM in US region is not able to reach LTM in Asia due to some network issue and reports it down but it is still marked up by Europe and Asia GTM as they still have connectivity .

 

Will all GTM will mark LTM VS down or just US GTM . And how will name resolution happen .

 

Thanks

 

8 Replies

  • I believe a monitor will only mark the pool member down on the local GTM, but when you administratively take a member offline it will propagate to all GTMs. I believe this is the reason prober pool exist. To probe from multiple locations.

     

    • SynACk_128568's avatar
      SynACk_128568
      Icon for Cirrostratus rankCirrostratus
      HI Brad , Thanks for the reply . So if one GTM marks LTM down it will not be propagated to other GTM's ? other GTM will respond to DNS query for that pool member ? And if it propagates will that be documented in /var/log/gtm of other GTM's ?
    • Brad_Parker's avatar
      Brad_Parker
      Icon for Cirrus rankCirrus
      I still haven't found a good way to test this as generally the iQuery and the sync traffic would share a network segment, so I haven't found a way to make the health check fail just for one GTM and still have connectivity to sync. My assumption is it would behave similar to LTM in that a failed monitor status doesn't get replicated since its technically not a configuration change. Try to test in a lab for a definitive answer.
  • I believe a monitor will only mark the pool member down on the local GTM, but when you administratively take a member offline it will propagate to all GTMs. I believe this is the reason prober pool exist. To probe from multiple locations.

     

    • SynACk_128568's avatar
      SynACk_128568
      Icon for Cirrostratus rankCirrostratus
      HI Brad , Thanks for the reply . So if one GTM marks LTM down it will not be propagated to other GTM's ? other GTM will respond to DNS query for that pool member ? And if it propagates will that be documented in /var/log/gtm of other GTM's ?
    • Brad_Parker_139's avatar
      Brad_Parker_139
      Icon for Nacreous rankNacreous
      I still haven't found a good way to test this as generally the iQuery and the sync traffic would share a network segment, so I haven't found a way to make the health check fail just for one GTM and still have connectivity to sync. My assumption is it would behave similar to LTM in that a failed monitor status doesn't get replicated since its technically not a configuration change. Try to test in a lab for a definitive answer.
  • Hi SynACk,

    the sync group implements a concept of delegating monitoring / health checking (and config distribution).

    This way a GTM typically monitors resources in the "local" data center and shares results with the other sync group members.

    That´s why you can expect a failing monitor to be propagated to the whole sync group in real-time.

    Just watch /var/log/gtm on all GTMs (
    tail -f /var/log/ltm
    ) to verify it.

    Btw, it is necessary that all GTMs are theoretically able to monitor all resources in your infrastructure. In case a "local" GTM becomes unavailable, another GTM takes the responsibility to monitor the related resources.

    Thanks, Stephan
  • Hi Techies

     

    can someone help with "GTM upgrade procedure" , currently we have four GTM two in UK & two in HK , all in global sync, i need step-by-step process and solution doc if available.

     

    Regards new techie