Correct Procedure to temporaily migrate vCenter VM

[Alan_ORiordan]

Hi

In my environment I have two live ESXi severs (VMWare Essentials license so no vMotion).

HOST1 (has the vcenter server VM runnning on it). I need to temporarily migrate this VM to the other local HOST so that I can run the vCenter Update Manager on HOST1 which requires all it's VM's powered down and put in maintenance mode. Being new to Veeam I plan to do the following:

1 - Create a replication job for the vcenter VM server which is a self-contained Windows 2008 R2 machine with SQL Express DB within. Should I use VSS quiescence? I have read this can cause problem's but that may have just been for backups?
2 - Power down the source VM
3 - Failover the Vcenter VM
4 - Confirm VM is working normally
5 - Remediate the patches on HOST1
6 - Failback to production the vcenter VM, once HOST1 has been updated and rebooted
7 - Confirm vcenter is working normally
8 - Commit failback
9 - Remove replica from disk within Veeam

I cannot use quick migration because my network label names are different which requires a re-mapping task.

[Shestakov]

Hi Alan,

Should I use VSS quiescence? I have read this can cause problem's but that may have just been for backups?

If you want application-consistent replication I recommend you to work with VSS quiescence.

Overall your plan should work. Just one thing: between steps 2 and 3 it makes sense to sync all latest changes to the replica VM.

Update: What version of Veeam B&R do you use? In older versions there were issues with vCenter backup; but, starting from v7 vCenter`s databases are excluded from VSS processing. More details are here.

Thank you.

[Alan_ORiordan]

Thanks Nikita, how do you mean "sync all latest changes to the replica VM" is this a veeam B&R task? Could you post a link to a KB or help article to explain this feature and what it does

Alan

[Alan_ORiordan]

We use Veeam B&R 7 with patch 4 installed

[Shestakov]

We use Veeam B&R 7 with patch 4 installed

Great. Should work.

Thanks Nikita, how do you mean "sync all latest changes to the replica VM" is this a veeam B&R task? Could you post a link to a KB or help article to explain this feature and what it does

There is no KB for that, probably because it`s not really difficult. Once VM is powered down, you just need to make the incremental replication job. It`s for synchronizing changes which were made in production site while the initial replication job(your step 1) was processing.

Thank you.

[Alan_ORiordan]

Ok, so if I amend the tasklist above, adding point 2.1:

1 - Create a replication job for the vcenter VM server which is a self-contained Windows 2008 R2 machine with SQL Express DB within, using VSS quiescence
2 - Power down the source VM
2.1 - Manually run the job again, to sync any incremental changes taken place during first sync
3 - Failover the Vcenter VM
4 - Confirm VM is working normally
5 - Remediate the patches on HOST1
6 - Failback to production the vcenter VM, once HOST1 has been updated and rebooted
7 - Confirm vcenter is working normally
8 - Commit failback
9 - Remove replica from disk within Veeam

[Shestakov]

Alan, that`s correct.

[Alan_ORiordan]

Thanks for confirming, the incremental job would only be required if the source was online when the replication job was sync'ing as no changes would be made if it was already powered down. I think I had the steps slightly out of order above.

Anyway, I will update with the results when complete for others reading this....

[Alan_ORiordan]

Once I had replicated and powered down the vcenter server I could not re-run the job or do the failover. I think this is because the only object I have added in the Backup Infrastructure > Managed Servers is the vcenter VM itself. That means that in the replication job it was referencing a machine that was then powered down to run the replication task or failover task. If that is the reason should all replication tasks be setup on the standalone hosts not the vcenter?

That also poses the question, (please excuse my novice experience level) of if the entire host is offline how would you complete the failover of the VM's that were on that host?

[Shestakov]

Alan,

The incremental job would only be required if the source was online when the replication job was sync'ing as no changes would be made if it was already powered down. I think I had the steps slightly out of order above.

Correct, if during replica creation the VM was not working and no changes appeared there, there is no need in another increment (step 2.1). However if the VM was working and you want those changes to be accomplished in the new VM, do incremental job.

If that is the reason should all replication tasks be setup on the standalone hosts not the vcenter?

Yes, for failover you need to add target host as standalone, because vcenter will be off in this case.

Thank you.

[foggy]

That also poses the question, (please excuse my novice experience level) of if the entire host is offline how would you complete the failover of the VM's that were on that host?

If your replication jobs are targeting vCenter, you can manually failover vCenter server before failing over other VMs through Veeam B&R console.

[Vitaliy S.]

....and in this case you will need to do all re-mapping tasks manually.

[Alan_ORiordan]

Would these re-mapping tasks be:

1) Power on the VM at the DR site.
2) Open the console view of the VM in vsphere and local in using local admin credentials because the IP would be wrong for the DR site.
3) Change the IPv4 TCP/IP settings to be valid for the new site.
4) Restart VM and check if DNS has updated on the local DR DC, if not, manually create the new DNS A record.
5) Open the DR Veeam B&R installation and then it will be able to communicate wth vcenter via the updated DNS records
6) Failover the other VM's

Should we want to conduct a DR test were the LIVE hosts were taken down, would we:

1) Power down the DR vCenter VM
2) Bring back online the LIVE Hosts and power on the LIVE vcenter machine on the LIVE host
3) Check and correct DNS, where required
4) Fail back to production the DR machines that had been used for the test
5) Re-check and monitor replication and backup jobs

In our case the vcenter server only contains vcenter and Veeam ONE so there should be minimal changes that would be lost simply by powering back on the LIVE vcenter/VONE server

[Vitaliy S.]

1) Power on the VM at the DR site.
2) Open the console view of the VM in vsphere and local in using local admin credentials because the IP would be wrong for the DR site.
3) Change the IPv4 TCP/IP settings to be valid for the new site.
4) Restart VM and check if DNS has updated on the local DR DC, if not, manually create the new DNS A record.
5) Open the DR Veeam B&R installation and then it will be able to communicate wth vcenter via the updated DNS records
6) Failover the other VM's

Correct.

1) Power down the DR vCenter VM
2) Bring back online the LIVE Hosts and power on the LIVE vcenter machine on the LIVE host
3) Check and correct DNS, where required
4) Fail back to production the DR machines that had been used for the test
5) Re-check and monitor replication and backup jobs

Correct, but next time I would recommend using a standalone connection to the DR ESXi host in order to automate the entire process.

[Alan_ORiordan]

Vitaliy

Hopefully one final question:

I want to update Host1 using vcenter update manager. If I successfully migrate the vcenter server to the HOST2 at the HQ site using the standalone host connection in the job and power it on could I simply just power off and remove the replicated copy of the machine and power back on the live machine and it would pickup the changes made to the host updates when re-scanned?

This would hopefully eliminate some of the manual tasks required to replicate back again such as removing VM's manually etc

[Vitaliy S.]

Yes, when you power on your live server back, I believe the connection to the host will be shown as "disconnected", so you will need to run a rescan job or reconnect the host to the vCenter Server. However, since there will be little changes done to the vCenter Server during host update procedure, I would suggest using failover/failback options for vCenter Server VM too.

[Alan_ORiordan]

Thanks all for your help over the last few days on this post. I am happy to say that I have managed to carry out this procedure now and I am happy all ESXi Servers are bang up to date.

Could I leave the vcenter replica on the other host indefinitely and just start a periodic replication of it instead of starting from scratch on that VM

[Vitaliy S.]

Yes, you can keep using that replicated vCenter Server as a target for the replication job. Glad to hear that you've successfully accomplished your update project.