Consistent Backup of vPostgres DB

[Nobody]

Hello Support

On vCenter 6.0, we will use the default vPostgres DB on our Windows vCenter.
What is Veeam suggesting to have a fully consistent backup of the DB without stoping any services?

Are there any plans to support the vPostgres DB directly?

Best regards
Andreas Girrbach

[dellock6]

Hi Andreas,
I'm not a PostgreSQL expert, but I was reading around for your same reasons, and found some info here: http://www.postgresql.org/docs/9.4/inte ... ackup.html

Looking at the three options, first of all file system level backup is not an option since it can only be done with the database in shutdown state. PG Dump seems to be a good solution, and looking at the commands seems it can be easily integrated with a pre-job script. This obsiously have the limit to restore the entire database at the same state of the time of the dump itself, but I think being it the database for vCenter, is good enough.
There's another option to do point-in-time backups which sounds similar to Oracle archive logs, but I don't honestly see the need to do so for the vCenter database, is more suited for other production uses of PostgreSQL.

Luca

[larry]

If you come up with a script please post it. Would be nice to have a Veeam best practice way of backing up the vCenter Server Appliance. Will be doing the update when the dust settles on 6.0.

[dellock6]

After a little bit of research, I've found VMware already has proper procedures and scripts available to backup the embedded postgreSQL of vCenter:
http://kb.vmware.com/kb/2091961
Just call the available scripts as pre-job activities in Veeam. One downside I see in the script is it creates a dump of the database thus consuming disk space, instead of simply preparing the database to be copied together with the entire VM.

[wRx7M]

I have successfully dumped the vcenter server vpostgres DB using the script but can't get it to run via the pre-job script section of my backup job. I have tried pointing to the bat on the VM guest (network share) and also on the Veeam server, locally. When I run the job, I get a warning "pre-job script terminated with exit code 1". What does this mean and how can I get it to work?

[foggy]

Ryan, can you check whether the script actually does what it is supposed to do when the job runs?

Also, here are some similar discussions, could help:
Pre / Post Job Scripts
Pre-job script terminated with exit
Is it possible to trigger a Veeam backup job from within

[Vitaliy S.]

You can trigger these scripts either via Veeam backup server or VMware Tools. Can you please check if triggering these scripts by enabling VMware Tools Quiescence option when creating VM snapshot solves this issue or not?

[giorgioC]

The PostgreSQL manual says that if the snapshot is a "consistent snapshot" of the data folder (and I assume VMware snapshots are consistent here), the backup will be able to recovery thanks to its journaling system (WAL logs). Now, assuming the application is doing its job (aka using transitions properly), a DB backup can be safely done without any pre-scripts. Do I have understood correctly?