Forum Discussion

Fremont's avatar
Fremont
Level 5
9 years ago

Best Practice for replicating EV drives for USL DR

The scenario is 2 clustered EV servers, EV1 cluster in PROD site, EV2 cluster in DR site.

For failover, customer is replicating index and storage locations.

Question: Do they also need to replicate the following locations from EV1 to EV2?

- msmq

- index metadata

- ev cache

- ev storage queue

- ev temp

 

The reason I ask is because, even though EV2 has no data of its own (strictly DR), however it has its own locations for the above 5.  Wouldn'r replicating cause corruption on EV2? 

For example, EV1 and EV2 are running their own msmq storage locations.  Replicating EV1's msmq will write over EV2's msmq storage location.  Would't that cause issues for EV2's msmq to start up?

Thanks,

 

  • AFAIK, There is no need to replicate / share following three location with EV2 in DR site.  

    - msmq
    - index metadata
    - ev temp

    Answers for following two location.

    - ev cache :- No need to replicate but location must be same (ex 'd:\evcache' should be same in primary & DR server).
    - ev storage queue:- It's should be on shared location that is accessible for both servers (ex '\\server\StorageQueue' location to be set in the properties of both the servers).

    Following Tech note will help you know additional requirement for building block configuration.
    http://www.symantec.com/docs/HOWTO109418

    I will also recommend to keep EV in backup mode during failover time due to one of the known issue listed in TN http://www.symantec.com/docs/TECH231820

    Recommendation for Building block with MSCS is mentioned in TN  http://www.symantec.com/docs/TECH215320

     

6 Replies

  • AFAIK, There is no need to replicate / share following three location with EV2 in DR site.  

    - msmq
    - index metadata
    - ev temp

    Answers for following two location.

    - ev cache :- No need to replicate but location must be same (ex 'd:\evcache' should be same in primary & DR server).
    - ev storage queue:- It's should be on shared location that is accessible for both servers (ex '\\server\StorageQueue' location to be set in the properties of both the servers).

    Following Tech note will help you know additional requirement for building block configuration.
    http://www.symantec.com/docs/HOWTO109418

    I will also recommend to keep EV in backup mode during failover time due to one of the known issue listed in TN http://www.symantec.com/docs/TECH231820

    Recommendation for Building block with MSCS is mentioned in TN  http://www.symantec.com/docs/TECH215320

     

  • couple comments and questions...

    why is the index metadata not required?

    if the storage queue should be on shared storage then it should be replicated.

    if a failover happens (unless planned for maintenance) how would he manage to get the system into backup mode first?

  • My thoughts are considering they are using two different EV clustered in different sites and on top of that they are using USL (Building blocks). May be we need some clarification the scenario from Fremont regarding cluster.

    Are you referring two clusters in different site, one cluster with couple of nodes is in primary site and second one is in DR site? OR Single cluster where nodes are spanned across multiple site?

  • To summarize my understanding, there is plan of building block failover between two clustered environment using USL. If that’s true then I would strict on my first comment however it may need to test out in lab environment to confirm. We don’t need to replicate IndexMetaData in different server instance as during initialization of Indexing service necessary meta data should create in background in DR server instance.

    Building block failover is always a planned activity & have to initiate DR process manually by changing DNS alias/run USL so EV vault stores can be placed in backup mode by using DR EV server instance. Keeping VS in backup mode is workaround till the defect (TECH231820) get fix.