Backup of a MariaDB Cluster

[GianlucaCroci]

Hello,

We've a cluster of 3 MariaDB servers on CentOS 7.
We'd like to know what's the best way to backup the Database. The important thing is to have consistency.

At the moment we only backup the VMs, and MariaDB every hour saves the Logs in a Storage outside the cluster.

Potentially we can install the Linux Agent, or better as we do now but adding Pre and Post scripts?

Thank's in advance for the replies.

[HannesK]

Hello,
just to be sure about the setup: it's about a Galera Cluster?

And what kind of restore is needed? Do you want to restore the database only, or the entire Linux machine (meaning you would try to join one old server from backup into a cluster that is still running on the two other nodes)?

My assumption is, that dumping the database is the best way, but let's see

Best regards,
Hannes

[GianlucaCroci]

Hello Hannes,

yes, it's a Galera Cluster.

We'd like to do any kind of recovery possible with database servers.
Then recover one or more Databases, as well as recover the entire Server.

With "dumping the database" do you mean through "Pre" and "Post" Scripts or directly from Galera/MariaDB on a partition that we would then go to backup with Veeam?

Thank's a lot.
Kind regards
Gianluca

[PetrM]

Hello,

With "dumping the database" do you mean through "Pre" and "Post" Scripts or directly from Galera/MariaDB on a partition that we would then go to backup with Veeam?

I believe this one is a reliable approach.

By the way, basic concepts outlined in this whitepaper are still actual. Also, you may have a look at this topic.

Thanks!

[GianlucaCroci]

Thank's a lot Petr

[GianlucaCroci]

Discussed with my colleagues who manage MariaDB's servers and DBs.
Unfortunately your proposals are not practicable for us.
First because the servers are always in use, and second because at the moment we're on a few GB of DB, but we expect growth up to a few TB, and we don't have enough disk space.

My proposal would be this, but I don't know how to implement it.
We backup externally to the cluster with CRONTAB, at the end of the Job, the active server should tell Veeam's Server to execute its Job (manual remote execution).
In Windows with powerShell I know it's possible, but on Linux?

Thanks.

[PetrM]

Hi Gianluca,

If pre-freeze/post-thaw scripts usage is not an option, then you may consider pre-job and post-job scripts. I suppose you can trigger a backup with CRONTAB from pre-job script.

Thanks!

[HannesK]

I see no chance for "entire server" restore because that we have no cluster awareness (I did not research whether it would be possible in theory)

I got lost between "and we don't have enough disk space." and "We backup externally to the cluster with CRONTAB". What does "backup" mean for you? Dumping the data to a file?

to the cluster

I would probably back up to a different machine. by doing that, the backup source for Veeam is always the same

scheduling the backups via crontab might be easier than using Veeam. If you use Veeam, you might need to increase timeouts as it takes some time to write TBs of data...

[GianlucaCroci]

OK, thank's a lot for the Help