Server Backup Methods

JohnB

How can we backup a server/node for disaster recovery? I'm not interested in doing a block level backup or an rsync of the whole system. Block level backups often cant be recovered in a VPS quickly and each provider has their own process. Rsync of a whole system leads to problematic restores.

What we need is a list of folders that enhance needs backed up so that we can do that and dump all the dbs.

From there, the recovery process would be install enhance normally on the replacement server with the same roles and then use rsync or whatever to restore the files/folders and restore the db dumps.

These are the obvious ones:
/etc/shadow
/etc/passwd
/var/local/enhance
/var/www/ client websites

@Adam could we please get some documentation on this so that we don't have to wait (or rely on) the promised disaster recovery feature. As it stands now, running in a public VPS means the only viable disaster recovery strategy is a script that recovers websites to a non enhance server.

Zoinkies

If it’s really critical for you, just wait?

JohnB

Zoinkies if you have already figured this out, I'm sure people would appreciate seeing your rsync solution.

As for waiting. Lol no. They will be late and the disaster recovery will rely on the backup server and not address how to backup the backup server. That's why I said "(or rely on)"

Adam

The server disaster recovery feature for web/db/email servers planned for May, may run into early June will rely on the Enhance backup system. The functionality will allow you to mark a server as "dead" and restore its sites to another server within your cluster from the last available backup, with all DNS (and CloudFlare if enabled) being updated automatically. This will also work with the upcoming S3 backup target.

We can potentially add a 3rd option in the future for customers who are not using Enhance's backup system. This would recreate the sites but without any of the website data, allowing you to restore it from your external backup system instead.

The probability of a complete server failure is very low and therefore I think a block level backup in addition to Enhance's website/email data backups is sufficient for most usage cases.

If you want to back up individual files, the strategy might look something like:

Rsync:

/var/local/enhance but excluding /var/local/enhance/containers
/var/www
/etc/passwd
/etc/shadow

Back up docker containers with "docker commit" and "docker save"

Back up docker volumes - you can probably directly back up /var/lib/docker/volumes

However restoring from these files into a consistent state is going to be difficult and is not recommended. Restoring a block level backup is likely to be faster.

Zoinkies

Adam with image/state backups, Enhance site backups, that’s enough. Even if you add in n-able or Acronis.

Powie

Adam sounds cool..... This what we are waiting for...

We have stopped the implementation of enhance because of this......

We know, its possible to do this via enhance support, but at this point we must have the ability that this can done by our tech staff at every time without external help.

looking forward 🙂

prasad0889

Would love to wait for the feature Enhance releases.

digitalexpanse

Q1
@Adam
Does this include the option to restore any site to an alternate server?

click, restore to X
this should be an option regardless of a server being marked dead

Q2
@Zoinkies where do these image/state backups come into play or have you dropped enhance backup and are using your own?

Adam

digitalexpanse Does this include the option to restore any site to an alternate server?

Yes, you can choose any server. It's going to work like the existing "move server" functionality except the data source is the backup rather than the existing server.