11-28-2019 03:16 PM - edited 01-17-2020 10:41 AM
Guys,
I have several linux VMs that are appliances. We cannot install agents on them as they are vendor locked and the only supported backup method we have is to power them off and run a vmware backup (either a vmware snap or Netbackup vmware backup). As such, how can I power off the vm prior to backup and power back on after? Can this be done natively in the policy in Netbackup? They do have vmware tools running (guest managed). If not, we can easily write a script to do this but where to run it (at the Netbackup master ?)
Environment:
11-29-2019 02:10 AM
May i suggest another method ?
Go back to the vendor of the locked appliances, and ask them how to protect the data the official way. Then the responsibility is on them, to provide a supported backup method.
12-03-2019 08:08 AM
I agree with @Nicolai this is how I've handled several different vendor appliances. They usually have a built in internal method of backup/recovery that requires that the appliance owner create backup config files and other things which could be dumped somewhere for you to backup. Usually, VMware backups aren't a supported method of protection for vendor specific appliances.
12-03-2019 11:53 AM
How many of those appliances come with VMware Tools installed already? If they are running VMware tools, you could try to back them up using a VIP which performs a snapshot and then the backup. You will not get the file-level restore ability but at least you can recover the VM from a backup image without having to redo the configuration.
12-04-2019 02:26 AM
Hi,
I did work on some projects that customers has uses some blackbox linux appliance systems and scheduled VM causes those systems to crash. I have checked that natively even doing snapshots via vCenter also cause the same issue therefore able to eliminiate that NetBackup causing the issues. Customers also have checked with their principle vendors on doing snapshot compatibility and vendors did replied as well as mentioned it is unsupported. Customers have mutually agreed to exclude these machines from backup.
As per suggested by @Nicolai, check with principle vendors on compatibility on doing snapshots backup as well as the native way of restoring those systems.
01-17-2020 10:47 AM - edited 01-17-2020 10:49 AM
@Nicolai wrote:May i suggest another method ?
Go back to the vendor of the locked appliances, and ask them how to protect the data the official way. Then the responsibility is on them, to provide a supported backup method.
Well, to be fair to the vendor(s), they do this. For example, Github enteprise is a Linux VM and the vendor offers a backup utlity here: https://github.com/github/backup-utils
That said, our estate is mainly windows and we backup everything to Netbackup. As we start introducing more linux appliances I wish to use the vendor native backup tools but I need to get them into Netbackup for various interal compliance reaons. I could for example in this case backup the file to an NFS share and have Netbackup pick it up I guess but then Netbackup is not directly responsible for that backup.
I think my question is a valid one.
01-17-2020 10:48 AM - edited 01-17-2020 10:48 AM
@X2 wrote:How many of those appliances come with VMware Tools installed already? If they are running VMware tools, you could try to back them up using a VIP which performs a snapshot and then the backup. You will not get the file-level restore ability but at least you can recover the VM from a backup image without having to redo the configuration.
All of mine have vmware tools but online vmware snapshot/restore not supported so not an option for me
01-17-2020 10:53 AM
Can I ask where you drove the scripts from ? Was it from a script associated with the master or an NB agent or completely external ? How did you handle script failures e.g. script fails to create the file so NB create 0 fiel backup from NFS share or the like? How did you alert on this condition?
02-04-2020 06:44 AM
Hi, we solved this problem recently, but our solution may not be appliable in your environment.
1. The linux appliance team scripts all applications to stop/unload at 02:00 AM.
2. In Vsphere we schedule the vm to "stop guest OS" at 02:30 AM.
3. Run a vmware backupjob at 03:00 AM using intelligent policy with condition that the vm is in a "power off" state.
4. Use Vsphere to schedule the vm to "power on" at 04:00 AM.
5. Linux appliance team scripts their applications to start at 05:00 AM.
I know all system can't be down for that long time during the night, we did not have that problem. You can probably run some tests to find out what time margins you can use.
03-06-2020 01:07 PM
Thanks for the info. I lik the intelligent policy that only runs if the system if in a powered off state. My issue is that I need a general pattern that can be re-used for this. My concern is that it seems above that this isn't really centralised and different workflows are running without direct knowledge of the state of other workflows. For example, the backup thats taken doesn't know if there were any erros during the shutdown phase within the application so who does it know its a valid backup form the NetBackup end? It doesn't. Thats not to say the admins don't know via their scripts.
I think I will look to automate via the new 8.2 API and the vmware API: https://sort.veritas.com/public/documents/nbu/8.2/windowsandunix/productguides/html/getting-started/