PostgreSQL Database HA

[eichenseherp]

Hallo,

wir haben folgende Backup-Struktur geplant:
Zwei Veeam Server die im Cluster laufen, wenn der eine Ausfällt wird das Config Backup am anderen restored und dieser übernimmt.

Wir haben geplant das die PostgreSQL Datenbank direkt auf dem Veeam Server läuft, wie ist es möglich, dass die Datenbank automatisch auf den anderen Server schwenkt? Cold stand by? Mit Patroni möglich oder gibts hier andere Lösungen?

Danke und Grüße

[Mildur]

Hi Patrick

This is an English only forum. Please continue this discussion in English.
To answer your question, no it's not possible. All credentials (application aware, vSphere, HyperV, Agent, ....) on the backup server are encrypted by the machine key of the first backup server. Besides other issues, the second backup server wouldn't be able to decrypt the credentials.

A supported way to migrate/connect the database to a new backup server is through a configuration restore (configuration backup must be encrypted).

What you could do:
1) Prepare a cold "stand by" backup server with it's own database.
2) create a configuration backup to a safe location in your secondary side
3) When you loose your primary backup server, import the configuration backup on your stand by backup server.

The entire process in step 3 will require you not more than 15-30 min.

We are aware of the request to provide some HA solution for our product, but today it is not available or possible.
Best option is to leverage a stand-by backup server or use virtualization for your backup server to be able to start it up on another virtualization host.

Best,
Fabian

[eichenseherp]

Okay thanks.
so there is currently no solution that the veeam database, which is located on postgresql, enables an automatic failover and switchover to the other postgresql database through third party providers such as patroni, or streaming prelication?

[eichenseherp]

Hi,
is there a sync/HA Featuere for Veeam in V12? Best would be an identical copy of Veeam Backup Server (incl. all jobs / configs. etc).
Or is there powershell / automation feature to restore the configuration backup automatically on site 2 and also do a copy job for veeam backup repos to site 2 and rescan them regularly?

The same feature for PostgreSQL Database available?

Thanks

[Mildur]

Hi Patrick

Please don't create additional topics for the exact same question.
I research the information yesterday with our teams.

- PostgreSQL Cluster for configuration database: may work for a single Veeam Backup server, but it is not supported. We run early tests on it and a failover to another PostgreSQL server node worked but required a reboot of the backup server (restart backup services).

- Backup Server HA: Not possible as explained in this topic yesterday. Use virtualization and shared storage to be able to power on the backup server on another hardware host. Use configuration backup to restore the configuration to a new standby backup server.

Or is there powershell / automation feature to restore the configuration backup automatically on site 2

With v12.1 we introduced a command line option to restore a configuration backup to a new backup server:

Veeam.Backup.Configuration.UnattendedRestore.exe /file:C:\backup\unattended.xml

You could write a script which restores the configuration backup. Please be aware, with a configuration restore the second backup server will take ownership over all repositories from the primary backup server. It's not a HA solution. It's a DR solution. Only use it when you have lost your primary backup server.

and also do a copy job for veeam backup repos to site 2 and rescan them regularly?

You can use backup copy jobs to copy backups to a repository in a secondary location. Rescan-VBREntity could scan an attached repository for new backups.

Maybe you can share more about the use case. Why is it important that a second backup server is ready to be used?
With zero downtime using a virtual backup server is the best option today until we may be able to deliver a HA solution in the future.

Best,
Fabian

[eichenseherp]

Thank you for your feedback.
We would integrate the two backup servers (hardware) in the failover cluster, with a shared storage where the config backup is stored. If the primary backup server fails, the config backup will be restored on the new one, where a dummy installation of Veeam is prepared. That would be the best or? https://www.veeam.com/blog/de/windows-s ... das_Backup

Because of the database. The goal would be to constantly synchronize the database on the primary backup server with the database on the cold standby backup server, so that the database is completely available on the second backup server in the event of a failure. Would Patroni be a solution here?

Can Linux repository servers, also here two hardware servers be created in the cluster?

Thanks!

[Mildur]

Hi Patrick

This article from Hannes is not related to any backup server components. It won't work for the backup server.

with a shared storage where the config backup is stored.

Why managing a windows failover cluster when it would be much easier to store the configuration backup on immutable object storage? Wasabi as an example is not expensive. Running a windows failover cluster must cost more over a year than using a Wasabi bucket for the configuration backup.

Because of the database. The goal would be to constantly synchronize the database on the primary backup server with the database on the cold standby backup server, so that the database is completely available on the second backup server in the event of a failure. Would Patroni be a solution here?

Thanks for explaining the use case.
Reality today: The second backup server won't be able to use encrypted data in the replicated database. That's already a blocker for your design (explained in my first comment). A PostgreSQL cluster may work, as long as the same backup server accesses the database. It cannot be a second backup server.

But as I said, we didn't tested this scenario completely and therefore we cannot support it. If you really want to use Patroni, then I suggest you set up a test environment and run your own tests. Our support team may decline any help if technical issues are related to the configuration database on your PostgreSQL cluster.
We know about the request to provide support for PostgreSQL clusters, but we are not there yet. It may come for a future version of Veeam Backup & Replication. Your request is noted.

Can Linux repository servers, also here two hardware servers be created in the cluster?

Maybe, but not tested by us. I'm sure you can replicate the storage on the operating system or storage level outside of Veeam Backup & Replication. And when the time comes, connect the second repository server to the backup server and import all backups. But our recommendation is to use a backup copy job so the backup server is aware of the restore points and can health checks on the backup files.

Today you must follow this approach:
1) Create backup jobs to a primary repository
2) Create backup copies to a secondary repository
3) Create an encrypted configuration backup to a secure location, if possible immutable to object storage
4) Have a stand-by server to be ready if the original backup server is lost
5) Import the encrypted configuration backup to the stand-by server when the primary backup server is lost. The stand-by backup server will be able to continue all backup jobs on your primary and secondary repository

I have noted all your points and your use case. It will be helpful to us if when we may start developing HA features in our product.

Thank you,
Fabian

[eichenseherp]

Hi,

thanks for the Feedback.
Can you tell us when Veeam is developing ha features?
Is there a roadmap?

Can we already prepare our system for the HA features, or can you not say anything about this yet?

[Gostev]

Yes, it really is too early to talk about this. The functionality should at least reach beta, before we can have meaningful conversations and preparations for it.