Support

ChatGPT 🤮 No, glorified autocomplete which needs hundreds of Watt-hours to come up with unreliable bollocks is definitely not the way to go about it. Let's instead use some power-efficient (33W TDP) hardware optimised for complex problem solving. Yeah, that's our brains right in our head.

As per the fine manual you can't really do what you want to do. UNiTE expects a filename. It cannot magically know your archive naming organisation and nomenclature which is necessary to determine what is the latest backup. So, we need to download that latest backup with a name UNiTE can recognise, and then tell UNiTE to restore it.

Therefore, we have these discrete steps:

1. Find my latest backup.
2. Download my latest backup.
3. Restore my latest backup.

Before we go ahead, let's note that since this is a matter of automation we want to use command line tools only. You could use AWS' own CLI but it's rather cumbersome since it's meant to work with all of AWS' services. Instead, we're gonna use s3cmd which is S3-specific, very easy to use, and very fast indeed. If you have not set it up already, go ahead and do it. I will assume it's accessible as s3cmd from your command line.

Find my latest backup

BACKUP_FILENAME=`s3cmd ls s3://my-bucket/site-www.example.com-* | grep .jpa | tail -1 | awk '{print $4}'`

This looks into the bucket my-bucket for any file whose name starts with site-www.example.com-* and prints them alphabetically. Since our naming convention is site-[HOST]-[DATE]-[TIME] alpha sorting is also time sorting.

In case we have multiple parts per backup we use grep .jpa to only fetch the .jpa files from our listing.

We then use tail -1 to only return the latest entry which looks like this:

2024-02-05 01:10    700550664  s3://my-bucket/site-www.example.com-20240204-190002cst.jps

Therefore we use awk to only return the fourth field, which is the filename.

Download my latest backup

s3cmd --skip-existing --no-check-md5 get $BACKUP_FILENAME

As simple as that. Now we get the latest backup file back from S3.

If it's a multipart archive we'll need to do a simple loop to iterate through extensions j01, j02 etc until we get an error.

Restore my latest backup

To restore the backup, we need it to have a predictable filename. So: 

mv $(basename $BACKUP_FILENAME) site.jpa

Now we can have our UNiTE file always try to restore site.jpa.

The attachments here are private. However, when you send me the XML file put it in a ZIP file, and replace real passwords and bucket names with dummy values. I just need to look at the structure of the file.

Your package key is wrong. When restoring from S3 you need to have

<package from="s3"></package>

This tells UNiTE to look at the <s3> section, download the archive part file(s), and use these as the file input.

This means that the bucket name is wrong.

Please remember that the bucket name is case sensitive. The bucket names FooBar, foobar, FOOBAR, and fooBar are different. Ideally, you should only ever use a bucket whose name is all lowercase.