This post is more than 5 years old
28 Posts
1
5812
Recovery Point Failover
Hi Guys,
This is jay, I need your Help on few things of Recovery Point.
I want do some Testing on a Consistency Group that is Critical and replicating to Other site.
We are going to do a Permanent Fail over in Future for this CG's, can we test by pausing the transfer of the CG and Mounting the Datastore at Secondary Site and see How it is working and the data in it.
I just want to do a Test on the CG before Main Event, I sthere any Method to do this without disturbing the Production.
Can we do a Test A Copy.
My Second Query is I have been getting some Alerts in third party monitoring tool solarwinds saying your Journals capacity is full, what happens the Journal's are full , do we loose any data?
Third Query: is I did Test a copy and Failover testing on one of the TEST CG, this is my first time, I just followed the procedure ,shutdown the source, bookmark, pause the transfer, mount the copy at secondary and it showed failover and I selected failover. If we want to fallback what is the procdure after failover , it didn't showed any fallback option.
If we were doing Permanent Failover at what time and step,weneed to Unmount the Source Datastore and when Mounting the Datastore at Secondary do we need to use the keep signature or new signature.
Kindly answer my Query's , that would be a great help.
echolaughmk
522 Posts
0
January 11th, 2017 11:00
You can do a "Test Copy" at any time once the CG is synchronized and has bookmarks to chose from. Doing the test will keep the prod online and provide you with a copy at the target site for you to play with and validate the data. If you are planning to do this right before the failover as a test (i.e - test and then immediately failover), then you can chose the "test a copy and failover" option in the GUI for that process.
for the second query - you journals will always be fully written after they ramp up storing all the snapshots. So that is normal to expect since they store all of the point-in-time images and fill up the journal and then rotate off as your protection window rotates.
for the third query - there is no "failback" at that point, you would essentially just repeat the failover process back to the original site and that would give you your "failback"
When you failover and execute that portion, that is when the source devices go WD so that is when you want to make sure all your applications are shutdown gracefully, etc.
jaiii
28 Posts
0
January 11th, 2017 11:00
When we do the Failover (Primary becomes secondary and Secondary Becomes Primary right), after the Failover when do we UNMOUNT the Secondary SITE Lun (DATASTORE).
At what point we need to unmount, can the both sites datastores be mounted ????????????????
Some Clearance needed in this Procedure.
jaiii
28 Posts
0
January 11th, 2017 14:00
What you are saying is Correct, Ialready did that, what iam asking is what happens when i finish the failover and my previous Source Luns are still in Mount Position and the VM associated is powered down.
Can the replication happen when the secondary Datastore and Primary datastore is in Mount Position.
Do we have any Impact if my Secondary(COPY) Datastore still in Mount after the failover happens and if the Virtual machine associated with it is Powered down.
ayman_shukry
1 Message
0
January 11th, 2017 14:00
Nice
echolaughmk
522 Posts
1
January 11th, 2017 14:00
When you failover in RP, you chose the image you want to bring up on the target side first. At that point the Source LUNs are online and RW and the Target Replicas are online and RW. The state on the target side is Logged Access.
Once you have validate the image you chose, using the wizard in RP, you then go back into the Recovery Activity where you will see your task waiting for you. At that point, when you chose to go back to the wizard and chose to "finish" the task, it will prompt you with the messages that data replication will stop and the journal will be lost - this is because you are starting the failover. It is prior to clicking this "Finish" button that you would ideally have all of your Source LUNs unmounted and inaccessible to the hosts at the source side because once you click this to initiate the failover, reverse replication will start automatically and what was your Source LUNs will now be your targets and be WD.
To go through the process, you can just setup a dummy CG with no data on it to replicate and test it out in your environment - that would give you an idea of the full process.
echolaughmk
522 Posts
0
January 12th, 2017 04:00
What is the ideal route is to not have your original source LUNs in the mount position with the VM's powered down. I would suggest to shutdown all the VM's, remove from inventory, etc prior to finishing the failover because if you leave them all there and mounted, but powered off, when the failover is finished, the LUNs will essentially look like they were ripped out from under the ESX hosts and all of your VM's will go to a grayed out, inaccessible state since it can no longer access those datastores (because those source LUNs are now write-disabled targets of the failover operation).
The replication can happen when both are mounted, but doing so will leave you in a funky state likely with some cleanup to be done on the original source because those LUNs will now be targets of replication in the reverse direction and will be write-disabled to the host, so none of the VM's will be able to write to them.