Video Screencast Help
Give us your opinion and win with Symantec! Please help us by taking this survey to tell us about your experience with Symantec Connect, so that we can continue to grow and improve.  Take the survey.

Symantec Veritas Cluster Server GCO query

Created: 09 Jun 2012 • Updated: 03 Aug 2012 | 3 comments
This issue has been solved. See solution.

Hi All,

The current VCS environment conists of a 2-node cluster at primary site and one node with VCS software at the DR site. The service groups were configured to provide failover for Oracle Apps and DB tier. VCS software was also configured at the DR site with one-node to provide failover for Oracle Apps and DB tier. However, when using GCO, we noticed that to bring up the database and the application service groups at the DR site, some amount of user inputs were required. This was actually the limitation from Oracle Apps - some SQL queries needed to be executed and the user was not comfortable performing the same. The user decided to remove the GCO componet and just name the server at the DR site with the same hostname of one of the cluster nodes at the primary site and then the database and application started working fine.

My question is related to license - Is VCS license required at all for the DR site? I know that this is not the recommended way of configuration, but can anyone please suggest what are the other disadvantages? I was looking at GCO and global service groups where a user cannot accidentaly bring up the application and DB at both the sites. We were using VCS agents for controlling NetApp SnapMirror replication which is also not possible now. 

Regards,

Bruce

Comments 3 CommentsJump to latest comment

Gaurav Sangamnerkar's picture

Hi Bruce,

Firstly, when you say that DR server has also been named with the same name of one of the server from primary site, to my knowledge this would not be recommended if you have used hostnames in the VVR configuration. Check vxprint -Pl output ... Now why I am saying above is:

VVR uses /etc/hosts file to resolve the name with IP address, if you make the same hostname for primary & DR, then /etc/hosts file will have issues providing different IP addresses ....

Secondly about license, I understand it is a one node cluster but still it can still be useful to bring the resources online following a proper dependency structure .. for e.g  bringing diskgroups / volumes / filesystems / NIC /IP etc online in proper order ... If you remove VCS license, things will work, but you are increasing the opportunity of human errors by making all above resources to come online manually .. the order of online would still need to follow ...

Thirdly, VCS would control the VVR agents too which can take care of site failover automatically ... now since if you remove VCS, you will need to take manual control on this ... one of the biggest risk which I believe is maintaining the flow of replication ... meaning, assume there is a site failure & you bring up the DR site ... that means your DR is now acting primary .. in this situation , replication should happen from this acting primary to old primary .. but I have seen many people confusing with this & replicating a wrong direction (from old primary to new primary) which may land up in incomplete data & finally landing in full replication once again ...

Hope this helps..

G

PS: If you are happy with the answer provided, please mark the post as solution. You can do so by clicking link "Mark as Solution" below the answer provided.
 

brucemcgill.nyc's picture

Hi Gaurav,

Thanks for the response.

We are using NetApp SnapMirror for storage replication. In the above mentioned scenario, VCS software cannot be configured with agents for SnapMirror and we are also not using Storage Foundation. This means that VCS cannot monitor the agents for snapmirror volumes. That will also be one big disadvantage. Please share your thoughts.

Thanks again.

Regards,

Bruce

Gaurav Sangamnerkar's picture

Hi Bruce,

sorry I missed the last line where you mentioned about Netapp snap mirror .... Agreed that would be a big disadvantage ... Its a very similar of case of VCS controlling any replication software ...

Lets take the example of Netapp snapmirror, should you have the VCS, it can monitors the replication state of filer devices, should a failover occur, the agent will reverse the direction of replication. These tasks would become manual for you if VCS is not there .. all the monitoring would go manual ..

Also, should any failure happen, the removal of lock files would be a manual operation. I believe without VCS there is a increase in chance of concurrency violation again because of lock file monitor.

G

PS: If you are happy with the answer provided, please mark the post as solution. You can do so by clicking link "Mark as Solution" below the answer provided.
 

SOLUTION