This content has been marked as final. Show 5 replies
MasterVIP takes two (V)IP addresses. As I understand it, the first VIP is assigned to the 'active' master and the second VIP is assigned to the 'standby' master. When there is a failover the VIPs move accordingly. This allows applications to explictly connecto to the 'active' or the 'standby'. Some apps may want to always connect to the standby (e.g. reporting apps).
Up on the topic.
Sorry for the delay, the platform to test on was not available to me...
I've done some tests using TimesTen with CRS, and I'm a bit disappointed : The VIPs supplied in the MasterVIP parameters do not seem to be switched when the roles of active-standby pair is switched !
If I use the first VIP given in MasterVIP, I always get on my "first" node datastore which can be either Master or Standby...
That leaves me with 2 questions :
- what VIP should the applications use in order to always connect to the master ???
- what is the point I'm missing with VIP that do not "switch" between datastores of the active standby pair ???
Hope someone can shed some light on this.
Self replying :
Ok, I think I got it.
In the case of an active-standby pair using clusterware, the client application should have TTC_server and TTC_server2 configured in the datastore declaration.
Something like :
Using upper masterDS definition, the client always get to the active datastore.
[masterDS] TTC_server=argument#1_of_masterVIP_parameter TTC_server2=argument#2_of_masterVIP_parameter TTC_SERVER_DSN=myDSN [standbyDS] TTC_server=argument#2_of_masterVIP_parameter TTC_server2=argument#1_of_masterVIP_parameter TTC_SERVER_DSN=myDSN
Using the lower standbyDS, the client always get to the standby datastore.