[Feat Req] Changes to SureBackup behavior

[ian0x0r]

Hi there,

I would like to have the option to allow a SureBackup job to continue running if it fails on a single VM when processing an Application Group. I know the Surebackup job will continue to run if its linked to a backup job.This is a long running debate as I came across a similar thread from 2012. vmware-vsphere-f24/surebackup-errors-an ... 14074.html

Is there any reason why the lab can not continue to run from the Application Group if 1 VM fails? Its pretty annoying, especially in the application group setup phase if 1 VM fails due to application test timeout. You have to wait a long time to then run the job again to test any changes to application test time out etc.

Thanks,

Ian

[k00laid]

My argument against this would be there are a number of ways to mitigate the issue with your VM failing due to application timeout while defining your application group. 1) you could increase the amount of hold time for that VM before tests are processed. 2) you can simply remove the application tests. In our environment the only test we leave is the basic ping test because we leave the labs up and let application users interact in it.

[foggy]

Agree. Besides, what is the reason to enable the test (or include the VM) in the application group if you do not bother whether the application actually works upon restore?

[dellock6]

I can answer for some very large customers I helped to in building large surebackup tests, where even 1000+ VMs where involved (https://www.virtualtothecore.com/en/can ... urebackup/). The premise is that it's simply impossible to do anything without automating surebackup like we did with those scripts, and application groups are the only ones that can be used to automate this process. And if we try to group VMs by 10-20 for example to limit the sprawl of jobs that need to be created, we want the job to complete anyway, and then tell me that X number of VMs have failed. I can then go in and fix those failed tests, as intended for surebackup indeed. But killing the entire job at VM #2 when there are other 18 to test, is honestly annoying.

[ian0x0r]

Hi Foggy,

For exactly the reasons Luca posted. Initial setup of a job with even 30 VMs that may need job time outs changing takes a very long time if every VM fails with application timeout. Imagine waiting an hour plus for cm 28 to boot only to find it needs its timeout changed. You then have to change timeout, start the job again, wait for an hour again and then hope it doesn't time out. Maybe a periodic check of tests would work. If I'm going to boot the application group for sandbox testing and leave it running, then. Yes it's nice of it passes the test but if it doesn't then I can examine th VM at least rather than Veeam just killingth job. I had a VM take longer than normal to boot on a test because it was pending a windows reboot. Manual intervention resolved this. Just looking for s halfway house between no tests and failing the job. If linked jobs do this, why can't application groups?

Thanks,

Ian

[foggy]

Ok, got it, thanks.

[mcz]

OK I just had a similar situation and checked veeam forums for an existing thread and here we are
The one thing is that surebackup stops when one vm fails and the other thing is that there is probably a way to boost up this thing a little bit...

I had to use a application timeout of 1800 seconds for my two DC's to make sure that the job won't fail. I am pretty sure that most of the time you could run the tests much earlier but you have to keep the timeout that high for those 3 times a month where the services won't be that fast up as normally (e.g. when the latest windows updates have been installed). I would say why not try to execute the scripts every 2 minutes? It won't need much ressources and if it fails, wait for the next 2 minutes until the timeout occurs. If it is successful after say 10 minutes, surebackup could start the next vm and would be faster in the end.

Probably you have even thought about this behaviour, maybe there are good reasons to not do it this way (allthough I can't see any at the moment)??

[PaulJGamma]

When creating a sure backup job, how about have an option to tell the job to just fire the VM's up in squence and not run any tests or use any time outs (other than one between VM's starting 20 or 30 seconds). Just to make sure they all do fire up, boot cleanly and have the correct IP information. Once you are happy thats the case you can then flip to test mode and you can then apply the scripts you need running and run through the real test as required. The option to turn off all the time outs and testing and to just run the sandbox would be very helpful.

Paul

[dellock6]

This last idea would, however, require to power on all the virtual machines first, and then run the tests against them. This would require a massive amount of CPU and memory, and create a lot of stress against our vPower technology. As much as I understand the reasoning behind this idea, I see it not being really feasible.

[PaulJGamma]

Well at that point it is up to the Admin setting up the Sure jobs to make sure there are enough resources to fire up the lab to run the test in the first place. That aside the option to have the sandbox with the VM's running with out the tests and time outs would just be a nice thing to have. In essance just turn the Sure Job into a straight up Sandbox option using the same functionality as the sure job with out the chance of a failed VM stopping the whole show.

[nicoSV]

Hi,

Our surebackup regroup many VMs in it, the problem is when one job has a failed status others job are skipped.
Is it possible to check those other job even if one fail instead of skipping others ?

If it's not possible i will have to make 1 surebackup for each VM and investigate to know why some jobs fail.

Have a nice day

[foggy]

Thanks for the feedback. This is possible if you link a backup job instead of using application group though.