Virtual lab questions

[jeremyh8]

We are wanting to create a virutal lab for a file server test. I am adding a dc and the file server to an application group. My question is do i need to allocate enough storage in my destination datastore for the entire content of the vms or as the manual says just redo logs? If it is just redo logs then this must mean it is booting the vms off of my backup storage and running it from there like an instant recovery correct? With that being said if i have a backup that has a retention of 7 days and I am runnign that vm in my virutal labe then what happens? Will backups even work or does that lab lock the backup file? What if i want to keep the lab up for 10 days does it extend the retention of the backup? I apologize for so many questions but i am having a difficult time grasping this for some reason.

[foggy]

If it is just redo logs then this must mean it is booting the vms off of my backup storage and running it from there like an instant recovery correct?

Correct. And all disk writes are redirected to the vPower NFS storage.

With that being said if i have a backup that has a retention of 7 days and I am runnign that vm in my virutal labe then what happens? Will backups even work or does that lab lock the backup file? What if i want to keep the lab up for 10 days does it extend the retention of the backup? I apologize for so many questions but i am having a difficult time grasping this for some reason.

Backup job will automatically stop any running SureBackup jobs which are locking the backup files that it needs to update. So, backup jobs will always complete fine. If you need to keep a lab for a long time, you could deploy a separate Veeam B&R server, copy backup files to it, import them there, and run required SureBackup jobs permanently.

[jeremyh8]

So for long term virtual labs i would need a different backup server only correct? I could still use the same storage which is attached to a physical proxy machine? On the physcial proxy i would just copy the last full backup to a folder called e:\virtlab\ then i would import that on a different b&r server and all should be good? i really apprecate your help!

[Vitaliy S.]

Yes, you got it correct.

[jeremyh8]

Well i tried the above scenario and i am having an issue.
1. Tried to start surebackup job. I get the following. Fail Error: Client error: File does not exist. File: [Bkp_Jax_pool0_daily_012012-08-14T180054.vbk].
Failed to restore file from local backup. VFS link: [summary.xml]. Target file: [MemFs://Tar2Text]. CHMOD mask: [0].
2. I verified that this is the backup file i imported.
3. It works without issue if i import the orignal bakckups but my copes it does not like. what am i doing wrong?

[Vitaliy S.]

Could you please send the entire SB debug log to our support team for further investigation?

[jeremyh8]

I ended up recopying the backups and it worked this time. well i atleast got farther! Still trying

[jg159357]

Good morning,

We have a group of developers working on our financial software. They currently have a development system that has gotten very out of sync with our live environment. We have a request to give them a live copy to do development on. Using sure backup I was able to create a virtual lab (with the isolation that lets me sleep at night), but it was only able to run until the next backup was taken. Is there a way to flag a restore as a long term recovery? If I do a vmotion of the "recovered" machines and power on the proxy machine would that work?

What are other people doing for creating labs of their live environments?

-Jeff

[dellock6]

Hi Jeff,
I usually create pre-emptively a small firewall appliance (I use pfSense, but any firewall distro is ok) acting as a double nat system just like the virtual lab appliance does, connected to a new dedicated portgroup.
Then, I simply restore all the needed VMs, give them a different name then the production copies, and after restore I change the network connection to the lab portgroup. If somebody needs their new VMs in no time, I simply restore them via instant recovery instead of normal restore, and then I storage vmotion them in a production storage.

Luca.

[jg159357]

Hi Jeff,
I usually create pre-emptively a small firewall appliance (I use pfSense, but any firewall distro is ok) acting as a double nat system just like the virtual lab appliance does, connected to a new dedicated portgroup.
Then, I simply restore all the needed VMs, give them a different name then the production copies, and after restore I change the network connection to the lab portgroup. If somebody needs their new VMs in no time, I simply restore them via instant recovery instead of normal restore, and then I storage vmotion them in a production storage.

Luca.

Do you create copies of your DCs as well? Do you give the restored machines access to the internet? How do you make sure they cannot talk to the real machines?

Correct me if I'm wrong, but this seems like a fair bit of extra work for something that Veeam already does fairly well for short time periods.

[dellock6]

You are right, but there are sometimes areas of the datacenter where I do not want to connect Veeam, and the mini firewall works even if I lost Veeam server.
And yes I restore DCs for testing so I can then simply delete them without writing anything on my production AD.
For real DR purposes, I usually have DCs replicated via AD, but I do not play at all with those....

Luca.

[Vitaliy S.]

Is there a way to flag a restore as a long term recovery? If I do a vmotion of the "recovered" machines and power on the proxy machine would that work?

No, that's not possible, so I would recommend to follow either Luca's advice or foggy's approach:

If you need to keep a lab for a long time, you could deploy a separate Veeam B&R server, copy backup files to it, import them there, and run required SureBackup jobs permanently.

[ekisner]

Is a surebackup job with the "leave running" checkbox checked supposed to terminate itself when a backup job runs on an associated VM?

If so, I'd like to submit that it should have the option of either mutexing out the jobs (within building the vlab for example) or even better, running it off of the existing backup without killing the vlab.

It was probably in the documentation and I just didn't see it, but sadly I just lost a day of testing after my surebackup job shut down the entire vlab overnight. My fault for not reading, but it does bring up an interesting possibility for a new feature (as described above).

[jg159357]

Yeah I was surprised the first time i saw it too. It does make a little sense depending on how your backups are being done; if you're running off of a full that the system thinks it needs to convert to a reverse backup what happens? Maybe a way to flag a certain backup each day/week/month as a lab source so the system could do preparation to use that file in the future?

I'll +1 for having an option to do a long-term lab from backups, but I have moved on to just doing a full restore for now.

[pkelly_sts]

Given a scenario where an entire set of VMs are replicated daily to a DR site, we know we can "put the replicas to work" by spinning them up at the DR site & doing testing etc.

What I'm not sure of is whether some of these replicas can be kept running (and still have their replicas updated) longer-term?

The problem I'm trying to solve is to keep a Dev version of 3 VMS (DC, App & SQL) running for, say, 1 month, after which time it'll be re-seeded/replace by a fresh copy of the live VMs. Logically, I'd say it can revert to a snapshot of some sort but I can't see how things would work between replciation snapshots.

Hey, if you don't ask you don't get

Paul

[foggy]

Paul, replication job will fail in case target (replica) VM is running at the moment when the job starts. Your scenario can be implemented with the help of additional monthly job or additional backup server and SureBackup/SureReplica functionality. Thanks.

[pkelly_sts]

Understood foggy, thanks for the response.

[Fizzer]

Hi All,
We've been using surebackup to provide our Devs with copies of their systems in an isolated environment so they can test upgrades.

Typically this means we've had surebackup jobs running for several days at a time. mostly this works OK, but the nature of surebackup is that it's designed to be transient so you can't run any of the source backups or reboot the vcenter server without the surebackup shutting down. We've had accidental destruction of surebackup environments before and we're not popular when it happens!

What we REALLY like about surebackup is the helper vm which allows us to redirect network traffic into the SB environment; allowing us to to rdp the test servers and do easy file transfers.

Has anyone set anything up similar to these helper VMs which can route traffic like that into at test environment? If so, what do you use? how do you configure it?

[foggy]

Andy, look through the answers above, should give you some hints. Thanks.

[Fizzer]

Thanks Foggy, I think pfsense might help us out here. Our problem with surebackup is it's reliance on the vcenter service.

Does anyone know of a good pfsense 'getting started' post or a config which might work for us in this instance.

Thnx

Andy

[foggy]

Does anyone know of a good pfsense 'getting started' post or a config which might work for us in this instance.

You'll probably find something useful here. Thanks.

[ekisner]

Hi folks, just finished setting up SureBackup. With a little help from support, so quick shout out to the ultra-fast response on a priority 4 ticket (within hours... they must've had a quiet day, which is always good to hear!). In case I wasn't specific, I've been using SureBackup for less than an hour now, so my experience and knowledge of its capabilities are effectively zero.

I have a question, and I figure the community would be able to answer faster.

I would like to have two Domain Controllers backed up with separate jobs, within an application group. Said separate jobs have staggered schedules, meaning that only one of the DCs will be getting backed up at any given time.

I would then like for them to stay running indefinitely. When a linked job runs, it doesn't have to start the application group because it's already started. When a DC gets backed up, after the backup is done SB just restarts the DC from the latest backup (with the other DC remaining online and accessible). I have no idea how well Active Directory will handle that (the offline controller coming online with a greater USN and such) but I think it should be fine, given how mature AD is as a technology.

Another feature I'd like to request is DHCP relaying. I'd love to be able to relay DHCP requests through vNICs. Not sure if it can do this already, I'll find out soon. It looks like the only option it has though is for a built-in DHCP server (which asks for DNS... perhaps it'll only assign the current A-record?).

I'm wondering if it's possible to have the DCs "leapfrogging" each other like that.

Learning, testing, liking.

[cramry]

I've thought about the surebackup job being able to be kept up and running as well for some of my virtual labs, would be a cool to know if that's possible. In all my experience trying, I can't get it to work, due to the surebackup job locking the application group's VM backup file while while it's running, therefore failing a backup job due to the file being locked and in use. Only way to unlock is to stop the surebackup job itself, which would then stop the lab. I just got upgraded to Enterprise + so I'm going to mess around with storage snapshots, maybe that could offer a solution, not sure though.

[ekisner]

I agree.. it would be useful for rapid provision of test environments as well. Rapidly deploy SB as a development environment, give developers access to the VMs in the lab resource pool, and let them go to town. In that case though, automatically restarting the VM with the latest backup may not be desirable.

Regardless, unless retention cycles are wanting to delete backup content that your lab is presently trying to use, I don't see why it can't just use the specific restore point, regardless of what the backups want to do (since it is indeed read-only).

[foggy]

Erik, do you mean SureBackup job finishes even in case the original backup job is using simple forward incremental backup method (so doesn't need write access to previous restore points used by SureBackup)?

PS According to the original thread, your experience and knowledge of SureBackup should be a bit deeper.

[ekisner]

Hah, well my "knowledge" and my "hey it would be cool if...." abilities are very different. I'm very good at asking for the moon. Knowing how to get it, however, very different

We presently use reverse incrementals, purely for the sake of faster restores, but it would be quite simple for us to switch over to regular incrementals.

The way I see it, (asking for the moon and what-not):
1) Backup job 1 for DC1 starts, DC1 in SureBackup Lab is killed
2) When the backup completes, DC1 powers back on in the lab using the latest restore point
3) Backup job 2 for DC2 starts, DC2 in SureBackup Lab is killed
4) When the backup completes, DC2 powers back on in the lab using the latest restore point
5) When a linked job completes, the SB job which is still running (as the lab has the DCs in the app group running) queues up the newly backed up VMs and does what it presently does (validate the backups by powering them on) then powers the linked VMs off per usual operation.