Testing Replica (B&R 6.0 and later)

[veremin]

Hi, Terrance, you have been merged to the existing discussion; so, kindly take a look at the answers provided above.

In general, since replica restore points are kept in VMware native format (snapshots), once the next replication cycle takes place, the changes made with VM you have been playing with will be discarded.

The description of replica manual testing procedure can be found here. Thanks.

[terrance101]

Hey v.Eremin, and tnx for putting me into the right area here, but just to confirm: I just brought up the replicated server on an isolated environment, created loads of docs on it, then shut it down and brought it back up but the docs were still there so shutting down did not remove them. However I know if I run the replication again from the live server they will go missing and the original server will be the replication again. Am I misunderstanding what you're saying here because it sounds like you said when the server is powered off on the replication all changes are lost?

[foggy]

This is exactly how you've described it, the changes made during the testing phase will be discarded on the next replication job run (corresponding delta VMDK file created during your testing will be deleted and new restore point will be created). That is how VMware snapshot tree works and that's also what Vladimir also meant in his post.

[veremin]

I’ve already changed slightly my original post in order not to cultivate further misunderstanding.

The point with testing replication is that once the next replication cycle takes place, the changes made will be discarded by reverting to original snapshot.

Hope it explains seen behavior better. Thanks.

[terrance101]

Perfect, thanks guys, maybe it was just the wording there that gave me a heart attack "powering down" and "losing all changes" eeeek, but now it's clear.

[veremin]

You’re welcome. Should any additional help be needed, feel free to contact us. Thanks.

[morciod]

We are planning on testing our DR strategy.
I need to know what is necessary for a test where we
stop replication
break our WAN link to DR site
start the replica VM's
shutdown replica VM
enable WAN link
restart replicaton

I have an OLDER documente that had us snap the replica prior to starting it.
i believe we got a message when VM was powered on where we had to reply: I copied it.

After test we removed snap and reenabled Veeam
We are running Veeam v6.5.0.144

Thank you!

[JosueM]

what I usually do is:

1. pause the replication jobs
2. take a before testing snapshot on replica vm to test
3. turn on the replica vm and do all the testing until finish
4. go back to prior testing snapshot
5. delete testing snapshot
6. resume replication.

[NightBird]

You don't need to take snapshot before starting replica since v6

[Gostev]

Correct, as VMware will create hidden protective snapshot automatically to enable going back to the state you've used ("Go To" functionality in the Snapshot Manager).

[murdocmk]

Unless you think that maybe during your restore testing you would like to "start over" due to some change you made, intentionally or accidentally, while trying to get things recovered. In which case, creating an additional snapshot isn't that bad of an idea.

[Gostev]

Start over is essentially "Go To" I was talking about... so, I don't really see how an extra snapshot would help. Unless I am missing some shrewd snapshot management scenario...

[murdocmk]

Yes, you're right .. duh, sorry!

[morciod]

can someone validate the process for exectuing a DR test. this is what i expect we need to do based on the doc i found- it just wasnt clear
1st we will disconnect network communications between source and replication site
- disable Veeam replication jobs
- snap replicatated VM ( we keep 2 snaps on replica)
- start VM

when test complete
- revert back from snap
- restart network communications
-reenable Veeam replication jobs

[foggy]

Debra, please review this topic for DR testing details. No need to create any snapshots.

[morciod]

how can i get to the replication v6 DR recovery docuemtation?

[Vitaliy S.]

Here are the instructions you should follow:

Q: I want to test replica failover. Can I simply power on any replica restore point with vSphere Client?
A: Yes. Because each restore point is a snapshot, its content will remain intact, because all disk changes inflicted by the running VM will go into the new snapshot file that host automatically creates to protect the selected snapshot's state (standard VMware snapshot tree functionality). For the same reason, Veeam replication job will not be affected either. Testing replicas with v6 is really a breeze - thanks to the new way of storing replica restore points.

[lemmy]

Hello,

We are running Veeam 7.0.1 in a ESXi 5.1 environment. Our setup is for disaster recovery, so we are replicating mission critical VMs to our DR site.

our VMs are replicating nicely, and we are getting ready to do some testing. Before we do, we'd like to load test our DR server.

Is there a problem in manually starting up the VMs to load test instead of doing a failover via Veeam? will this cause any problems?

[veremin]

Hi, Lemmy,

Your post has been merged into existing discussion regarding replica testing. In order to test replica VMs you can either use procedure described in this thread and in our sticky FAQ, or utilize SureReplica functionality that has been introduced in version 7 and that allows you to test replica VMs in automatic fashion.

Thanks.

[frigomiam]

Hi all,

I've got my DRSite with vsphere and R&B v7.0,
Replica VMs can't be powered up at as it is as they are on same network as prod VMs.

My veeam R&B server is in my prod.

I used to test my DRSite this way
- A / create snapshot on replica VMs + change network of replica VM to internal network ( not linked to prod ), power on replica VM
OR
- B / cutting network between DR and Prod + create snapshot on replica VMs + power on replica VMs

I see that normally the failover/undo failover functions are used in R&B for such purpose.
So is this function possible only for replica VMs not connected to prod network.

creating snapshot and reverting manual is a bit long, and prone to errors.

Does it make sense that I do the following in plan B/ :
- restore a veeam on DRSite using configuration backup
- use the failover function from this new veeam restored veeam ?
- or is it too much hassle and I should I stick to the manual way.

In a real DR scenarion, I should anyway restore veeam too right, then failover the VMs using the restored veeam server ?

thanks

[foggy]

creating snapshot and reverting manual is a bit long, and prone to errors.

There's actually no need to create snapshots, you can just start the required replica VM restore point. All the changes will be discarded during the next job run.

In a real DR scenarion, I should anyway restore veeam too right, then failover the VMs using the restored veeam server ?

Correct. Another approach is to have a second Veeam B&R instance responsible for replication jobs located in the DR site, this will allow for immediate automatic failover of all VMs.

Consider also using SureReplica for automatic testing of replica VMs.

[aeccles]

There's actually no need to create snapshots, you can just start the required replica VM restore point. All the changes will be discarded during the next job run.


I've spent a bit of time trying to figure out the best procedure for manually testing replica's and the came across this topic which states numerous times that you don't need to create a snapshot before testing and that any changes will be discarded during the next replica job. That is great, but I had opened a case earlier (#01019829) and was told this:

VMWare doesn't automatically create a snapshot when you power on the VM. Its actually recommended to take a manual snapshot of the Replica before powering it on for testing. Then you revert back to that snapshot once you are done testing. The reasoning behind this is: When you re-enable the connection between your Prod and DR site, and re-enable your Veeam job, the Veeam job expects there to be no changes between the two VMs. It can create an issue if the Production VM doesn't match the Replica VM that was altered. It won't break it, but several senior engineers advised me its a good rule of thumb just to do the snapshot from the beginning just to avoid change data from having to be sent over.

We do have a KB on how to manually test a replica via the vSphere client, without affecting the production environment. It does not list the pre-powering on the Replica Snapshot, but its still a good idea. But this explains why you don't see a snap being made. Because its not automatically done by VMWare, its manually done by you.

http://www.veeam.com/kb1073"

So, it seems like it is not necessary but their might be a performance benefit during the next job? Is that correct ?

[Vitaliy S.]

aeccles, thanks for posting back. As per the KB you've mentioned, you do not need to create snapshot manually prior testing the VM replica, and there is no performance benefit of doing that. It was the case for previous versions of Veeam B&R (manual VM snapshot creation), so the recommendation from the support engineer is a bit outdated.

[james411]

Hi,

I read the FAQ, but I wanted to confirm my procedure before testing this as it always makes me a little nervous

I have a replica job running every 15 minutes for a SQL and Exchange Server. I want to test the replicas and make sure everything works. Would this be correct:
1) Disable replica job (since it runs so often, I don't want it to run while I am testing)
2) Change replica vm settings to move nic to isolated vSwitch
3) Power on replica from vSphere client and test
4) Shut down replica when done testing
5) Re-enable replica job

Questions: once done, do I need to change the replica VM's nic back to what it was before I changed it to an isolated vSwitch? And are any changes I make while testing kept with that replica restore point?
Running Veeam BR v8 U3

Thanks!

[PTide]

Hi,

Have you considered using SureReplica job instead of manual testing? The steps you've described are correct, however I think it'd be more convenient and easy to utilize SureReplica job in order to check you replicas' recoverability.

Thank you.

[foggy]

Questions: once done, do I need to change the replica VM's nic back to what it was before I changed it to an isolated vSwitch? And are any changes I make while testing kept with that replica restore point?

Any changes performed during replica testing will be discarded during the next job run.

[james411]

Have you considered using SureReplica job instead of manual testing? The steps you've described are correct, however I think it'd be more convenient and easy to utilize SureReplica job in order to check you replicas' recoverability.

Thanks for the confirmation. Unfortunately, I'm only licensed for the Veeam Essentials kit so no SureReplica for me

[nunciate]

I need to test something on a very large file server and I want to use my DR replica to do the testing.

Is it safe to manually create a snapshot on a replica which already has 2 restore points.
First I would create the snapshot on the replica VM (on top of the restore points).
Then I would power on the replica (manually or maybe with Failover Now in Veeam).
I would give the server a new name and IP and bring it back onto the network for testing.
Once testing is completed I may consider leaving the replica in DR up permanently with the new name.
I might also need to revert the VM to the snapshot I created if I am not going to keep the changes.

I am assuming this is going to break CBT and require disk recalculation but I was wondering if I should be concerned about any other issues or corruption of the replica.

[foggy]

Hi Alan, you do not need to make additional snapshots to test replica VMs. They are already protected by the snapshot and all changes will be discarded if you decide to power the VM off and continue replication. Please see above for details.

[blan]

Hi all,

Hoping someone can lead me in the right direction with my scenario. I currently just upgraded to Veeam 9 running at our main site which is replicating successfully to a DR location. The sites are connected via a site to site VPN, however both sites are on the same network subnet.

What I'm trying to achieve is a DR test with the replicas. In order to do a valid DR test I'd like to be able to be at the main site and remote in using RDP or any other remote connection method and access those replicas.

1. What I'm not sure about it how can I power up this replicas since the DR site is on the same subnet?
2. I've been reading up how to use Re-IP, but I don't see how that would work because if we go and change the IPs of domain controllers, Exchange servers, SQL servers, etc doesn't that leave the possibility of many other issues happening, such as DNS issues?

Any ideas or thoughts are highly appreciated.

- bl