Replication reverse

If you plan on returning changes on the destination side back to the source, then you will need to reverse.  If you're just doing a test and will discard any changes made on the destination side, there is no need for a reverse.

Reverse = changes the direction of replication.

Switchover = Syncs any outstanding changes and then switches R/O / R/W status of the replicas.

Failover = Switches status WITHOUT syncing changes.  Useful during an actual failure.

The plan is to switch it over and leave it on the new array as PRODUCTION.

The old array that (at that point) was production is now a REPLICATION TARGET for the new array.

The reason I threw the question out there is because I was talking to a person who was supposed to be a NAS authority and they were telling me to do a "REVERSE" and not a Switchover, which didnt sound correct to me and caused me to doubt my thought process.

I think this would be the correct order in what to do.

1. Switch over VDM
2. Switch over File Systems
3. Change IP addresses in DNS (target array IP addresses are different)
4. Stop replication
5. Reverse replication
6. Start replication

Go grab a beer!

The difference are subtle and are often confused – people use failover when they mean switchover and vice versa

Just read what the manual says and choose the behavior you want

Rainer

If your CIFS servers automatically update DNS, there should be no DNS change necessary; DNS should be automatically updated.  If you have disabled automatic updates, then the DNS change will be necessary. 

Also, Reverse is a single operation.  It doesn't require a start/stop.

I would totally use 'Reverse' in this example... 'Switchover' does not keep replication going the other way. It stops replication after doing a pretty complex list of steps. If you wanted to move to the new array and shut down the old one, switchover is your friend.

Now, don’t get me wrong, if you use switchover, it is still possible to start replication backwards. You will need to use a ‘start’ command with a ‘-reverse’ option, but my experience shows that people get very confused with that, especially when they read the man pages…

So, if you want complexity – go with switchover, followed by ‘start –reverse’. If you are just swapping which array is production and which on is DR, and want to do everything synchronously, use ‘reverse’.

HTH,

--MR

When doing reverse operation ckpt_schedules will move to new source array? bcz i observed in one of the case that ckpt, F/S were moved to new source but ckpt_schedule stays back

Hello vignesh21,

Which Celerra system do you have and what is the current version of firmware that is running on the system?

It is a eNAS array and current code version is 8.1.15.323

Hello vignesh21,

No the ckpt_schedules will not move to the new source array.

Hi,

Can you explain why the ckpt_schedules not moving from one array to other during reverse syncrep operation. I have tried to perform reverse syncrep in my lab setup[code version 8.1.12-313] and observed that ckpt_schedules were moved from old source to new source array[from which i executed reverse syncrep command]

Hello vignesh21,

Sorry I was wrong as I was reviewing an older document then your version you are running.  The schedule should transfer over.  In cases when it doesn’t there is normally a space issue or a permission issue that is the cause for it to not transfer back over.