Question on usage of Replicator V2 (for DR Test)

Are you going to be writing to DR file systems?

Sent from my Verizon Wireless 4GLTE Phone

If you are going to discard writes after the test you dont even have to stop replication, just create a writable checkpoint and export that via nfs

Yes, the file systems will be mounted on the DR servers and writes will take place. (this is to test. Just to make sure how it would work in a real DR situation)

I am sorry.  Our Production is going to be Up (just got the news). We are going to just stop the replication. Make all the File systems in DR as RW, perform test. Change the file systems to RO again and start the replication again.

We have 40 file systems replicating. Stopping each of them and setting them to RW would take long time. (if we do it one at a time)

What would be the best approach ?

stopping replication will not make destination RW.  If you have no plans of keeping changes that were made to DR file systems during the test, i would not even bother stopping replication. Create a RW checkpoint on DR Celerra and export that to your DR servers. When all tests are done you simply discard those RW checkpoints and go your marry way ..no need to mess with production. Take a look at this solution, this is for CIFS but you get the point.

emc216579

as far as automating this, you have bash shell on Control Station,  a couple of "for loops" will get the job done pretty fast

Before automating I would suggest to first understand

The Replicator manual might not be the most exciting read but it does explain the differences between failover, switchover, stopping, …

Rainer

why do you even want to stop replication ?

For testing, both prod and DR will be up at the same time.  We will not be using DR as the prod instance and replicating any changes back. We have all the file systems from the source at the target site. There are two applications for which replication is set.

Application 1 uses

Data_Mover2 on Source  --> Data_Mover2 on Target

Application 2 uses                 

Data_Mover4 on Source  --> Data_Mover4 on Target

We have DR for application 2. Two different Interconnects are setup for different data movers above. I am finding it difficult to stop the replication for file systems to specific data movers. nas_replicate does not have many options. Unlike, fs_replicate - eg: fs_replicate -suspend and fs_replicate server_2 -list

nas_replicate

   -list [ -id ]

$ nas_replicate -stop { | id= } -mode {both}

I want to know we can pause the Interconnect. I read it in the Using Celerra Replicator (V2) document that

"When you pause a Data Mover interconnect it temporarily stops all data transmission for all replication sessions using the specified interconnect. Data transmission remains paused until you either resume transmission over the interconnect"

$ nas_cel -interconnect -pause { | id= }

Is this a good option ?

i've been proposing that for 2 days now

There is no need to stop replication.  Your easiest solution is to create writable checkpoints on your destination Celerra and mount those to your servers.  Your servers can test the data using those, and when the test is done, you simply delete the writable checkpoints.  There is nothing more you need to do.  No need to stop replication, no need to suspend the link, etc.

I totally agree. But, the business is treating this as the real DR and wants to stop replication. (i am helpless)

Last DR was done way back in 2009. The SOP's are outdated ( I am re-doing the SOP).

OLD SOP  (Begin DR)

Step 1  Suspend IP Replication ( fs_replicate -suspend )  from source.

Step 2  Convert filesystems to UXFS  (nas_fs -Type uxfs -Force)

Step 3  Create a consistency checkpoint  (fs_ckpt -name _consistency -Create )

Step 4  Remount as RW  (server_mount server_2 -o rw )

OLD SOP (Undo DR)

Step 1 Remount all filesystems as R/O   ( server_mount server_2 -o ro )

Step 2 Roll back to consistency state  (/nas/sbin/rootfs_ckpt _consistency -name _restore -Restore -o automount=no )

Step 3 Convert filesystems back to RAWFS

Step 4 Unmount and Delete consistency checkpoints

Hence, I was looking for the info if pausing the interconnect would stop all data transmission for all replication sessions using the specified interconnect.

If you want to test a real DR then I suggest to follow the procedure in the manual for DR

For finalizing the procedure I would suggest to setup two VNX simulators and try it with them before the real environment.

Rainer

Thank you for your suggestion. It helps a lot. I think the DR test will  not last for more than 48 hours (2 days).

(nas_replicate -stop rep_fs01 -mode both ) - this will stop a single instance. I have over 30 FS replicating. I will have to create a script to stop all of them.

nas_cel -interconnect -pause { | id= }   - will stopping the interconnect do the same thing ?

if they insist on stopping replication, sure ..go ahead and stop it (nas_replicate -stop rep_fs01 -mode both), at this point file system is still read-only. Create a writable checkpoint and give that to your customer. When testing is complete, delete writable checkpoint and restart replication (nas_replicate -start rep_fs01).  How long do you think they will be testing on DR side ? While replication is stopped, savvol on the source is filling up, if it reaches certain threshold, you will not be able to restart replication and will have to do a full sync. (that's why we recommend not to stop replication at all)