3.5 to 4.0 software upgrade -- 10 GB min. jvols

Yes, you can stop all application I/O, issue a bookmark for all the CGs, check that the bookmarks have replicated to their target copies, reconstruct the CGs with the new journal sizing, enable the CGs but do not start transfer, then use the clear markers option via the CLI for each CG/copy and finally start transfer for all the CGs/copies. Once this has been done you can start application I/O followed by a image access test to verify data integrity by utilisaing the PiT rollback from the journal.

Thank you forshr.  I'm not fluent in RecoverPoint, obviously, so I don't understand everything you're recommending.

All our Remote journals exceed the 10 GB requirement, and that's the important thing for DR.  It is only our Production journals that are too small. 

Will this approach work only when all CG I/O is totally stopped, because in our setting I think that would mean taking a host or cluster down (some of this data is boot devices, so stopping I/O is impossible).  The best I can do is plan for a period of minimal I/O.  Is that good enough? 

If so, here's what I think you're recommending.  Please correct me where I'm off-base, which I will frequently will be.

1.  During a period of minimal I/O (the best we can do here) "issue a bookmark for all the CGs"

2.  Check that the bookmarks exist (using the RPA GUI in my case). 

3.  Then you write "reconstruct the CGs with the new journal sizing".  I'm lost.  Do you mean on the VNX Expand the existing journal LUNs in instances where they are less than 10 GB minimum?  Or do you mean re-do the CG entirely, with new 10 GB journal LUNs?  Or do you mean something else entirely?        

4. Then you write "enable the CGs but do not start transfer".  You didn't mention disabling the CGs anywhere, so I'm lost again!

5.  Then you write "use the clear markers option via the CLI for each CG/copy"  That's the "clear_markers" command?  I'll read more about that.  I notice that it exists in the GUI as well.

6.  Then you write "finally start transfer"

If you could explain and/or expand on this I'd really appreciate it. 

Thank you.

JimP

Hi Jim,

Unfortunately the 10GB journal requirement is for all copies journals in the CG. You need to cease I/O to ensure that the source and replica volumes are identical in conjunction with the clear markers command (sets all the markers for all the volumes as identical and avoids the full sweep) which exists in the UI at 3.* but is only a CLI option at 4.*.

Regards,

Rich

Hi again Rich.

You write "cease I/O."  I can't think of any way to do that, short of powering down the hosts.  Is there something else you have in mind, I hope?  Shutdown fiber links to the RPAs while doing the CG reconfiguration?

I'm struggling to find a way to avoid a full sync, which will take weeks.

Thank you.

Hi Rich.

Thank you for the quick reply.  I'm off on another tangent.   

On page 205 of the RP 3.5 Admin Guide there's a heading entitled "How to change the storage capacity of a journal by changing the storage capacity of a journal LUN on storage."  Would this procedure work?

My reading of it is that the journal is entirely removed is step 1e.  In step 2 a new larger LUN could be created on the array.  In step 3 a New Journal Volume is added.

Is this a blind alley?

Thank you for your help.

Hi Jim,

So here's the rub. When you use the clear_markers command you are telling RP that the source and replica volumes are identical and therefore avoiding the need for the full sweep whereby blocks are compared and differential data is replicated. In other words it creates a identical bitmap for every volume per CG. The source and replica cannot be identical if you do not cease I/O to the source volumes.

Regards,

Rich

And hello again.

In the meantime I've sort of tested the procedure "How to change the storage capacity of a journal by changing the storage capacity of a journal LUN on storage" on page 205 of the 3.5 Admin Guide", up to step 1.c. 

Step 1.c is:  "Select the group that the journal volume belongs to in the Navigation Pane, select the Replication Sets Tab, and double-click on one of the replication sets in the replication set list.  The VOLUMES CONFIGURATION dialog box is displayed..."

However, when I "double-click on one of the replication sets in the replication set list", the REPLICATION SETS PROPERTIES is displayed, not the VOLUMES CONFIGURATION dialog box.

What am I missing?

Thank you.